A critical review of the epidemiology of Agent Orange or 2,3,7,8-tetrachlorodibenzo-p-dioxin and lymphoid malignancies

Annals of Epidemiology 25 (2015) 275e292

Contents lists available at ScienceDirect

Annals of Epidemiology journal homepage: www.annalsofepidemiology.org

Review article

A critical review of the epidemiology of Agent Orange or 2,3,7,8-tetrachlorodibenzo-p-dioxin and lymphoid malignancies Ellen T. Chang ScD a, b, *, Paolo Boffetta MD, MPH c, Hans-Olov Adami MD, PhD d, Jack S. Mandel PhD, MPH a a

Health Sciences Practice, Exponent, Inc., Menlo Park, CA Division of Epidemiology, Department of Health Research and Policy, Stanford University School of Medicine, Stanford, CA c Institute for Translational Epidemiology and the Tisch Cancer Institute, Icahn School of Medicine at Mount Sinai, New York, NY d Department of Epidemiology, Harvard School of Public Health, Boston, MA b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 19 September 2014 Accepted 9 January 2015 Available online 30 January 2015

Purpose: Establishing a causal relationship between 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) and risk of specific lymphoid cancers, including non-Hodgkin lymphoma (NHL), Hodgkin lymphoma (HL), and multiple myeloma (MM), would be useful for risk assessment. Methods: This article systematically and critically reviews epidemiologic studies of the association between exposure to TCDD or TCDD-contaminated herbicides and risk of lymphoid malignancies. These include studies of military, industrial, accidental environmental, and general environmental exposure to Agent Orange or TCDD. Results: Collectively, the epidemiologic evidence from industrial cohorts suggests a positive association with NHL mortality, but results are not consistent across other studies, a clear exposure-response gradient is not evident, and data are insufficient to conclude that the association is causal. Furthermore, available studies provide little information on NHL incidence or specific NHL subtypes. Epidemiologic studies do not show an association of TCDD exposure with HL, whereas the indication of a positive association with MM in a limited number of studies, but not others, remains to be confirmed in additional research. Exposure classification error and small numbers are important limitations of the available epidemiologic studies. Conclusions: Overall, a causal effect of TCDD on NHL, HL, MM, or subtypes of these lymphoid malignancies has not been established. Ó 2015 Elsevier Inc. All rights reserved.

Keywords: Lymphoma Lymphoma, Non-Hodgkin Lymphoma, Hodgkin Multiple myeloma Tetrachlorodibenzodioxin Agent Orange Epidemiology

Introduction Lymphoid malignancies, which are classified broadly as nonHodgkin lymphoma (NHL), Hodgkin lymphoma (HL), and multiple myeloma (MM), comprise a heterogeneous group of clonal tumors of mature B cells, T cells, and natural killer cells [1]. Although occupational exposure to pesticides in general has long been studied as a risk factor for NHL [2e4], HL [5,6], and MM [7e9], no causal association has yet been established with any specific pesticide. Agent Orange, an herbicide used by the U.S. military for strategic defoliation during the Vietnam War in the period 1962 to 1971 [10], was contaminated by 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). * Corresponding author. Health Sciences Practice, Exponent, Inc., 149 Commonwealth Drive, Menlo Park, CA 94025. Tel.: þ1-650-688-6734; fax: þ1-650-6881799. E-mail address: [email protected] (E.T. Chang). http://dx.doi.org/10.1016/j.annepidem.2015.01.002 1047-2797/Ó 2015 Elsevier Inc. All rights reserved.

Since 1991, the U.S. Institute of Medicine Committee to Review the Health Effects in Vietnam Veterans of Exposure to Herbicides (IOM Committee) has conducted comprehensive reviews and evaluations of the available scientific and medical evidence on associations between specific health outcomes and exposure to Agent Orange and other herbicides used by the military in Vietnam. In each of these biennial reviews, the IOM Committee has determined that “evidence is sufficient to conclude that there is a positive association” between herbicides and both NHL and HL, such that “chance, bias, and confounding could be ruled out with reasonable confidence” [10e19]. For MM, the IOM Committee considers the evidence of an association with herbicides to be “limited or suggestive.” Of note, the IOM Committee is charged specifically with determining “whether a statistical association with herbicide exposure exists, taking into account the strength of the scientific evidence and the appropriateness of the statistical and epidemiologic methods used to detect the association” (emphasis added) [10]. The

276

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

IOM Committee “fully recognize(s) that an association does not establish a causal relationship and that the rigor of the evidentiary database needed to support a finding of statistical association is weaker than that needed to establish causality.” Moreover, any association is not necessarily specific to Agent Orange or other herbicides that contain TCDD. The epidemiologic evidence on Agent Orange or TCDD and risk of lymphoid malignancies is complex. A detailed examination of this relationship is necessary for informed risk assessment and public health regulatory decision making. Although high-level exposure to TCDD and other polychlorinated dibenzo-p-dioxins is rare in general populations, due in large part to restrictions on the manufacture and use of chlorinated compounds, low-level exposure to dioxins is widespread, presumably through diet [20]. Moreover, the results on HL and MM have not been subject to recent systematic reviews, and the potential adverse health effects of Agent Orange exposure in Vietnam continue to be a focus of public concern and controversy [21e24]. Therefore, to clarify the current state of the science, this article systematically reviews the epidemiologic evidence on Agent Orange or TCDD exposure and risk of lymphoid malignancies. Epidemiology of lymphoid malignancies As the 10th most common malignancy worldwide, NHL is diagnosed in approximately 386,000 new patients and causes approximately 200,000 deaths each year [25]. MM and HL are substantially less common, with the former affecting approximately 114,000 new patients and causing approximately 80,000 deaths annually, and the latter being responsible for 66,000 new cases and 25,000 deaths annually. However, half of incident HL cases occur between ages 15 and 44 years, making HL the 11th most common malignancy in that age group worldwide and the sixth most common in economically more developed regions [25]. NHL incidence rates vary by more than 20-fold internationally, and HL and MM incidence rates vary by more than 30-fold, with generally higher rates of all three lymphoid malignancies in Israel, North America, Australia and New Zealand, and northern and western Europe and lower rates in Africa and East and Southeast Asia [25]. However, certain NHL subtypes, such as Burkitt lymphoma and adult T-cell leukemia/lymphoma, are more common in the latter regions [26,27]. In most countries with cancer registries, the incidence rate of NHL increased during the second half of the 20th century, followed by a plateau in many regions [28]. The incidence rate of MM also followed the same pattern in several geographic areas. These trends are believed to be at least partially explained by improvements in case ascertainment, reporting, and diagnosis [29,30]. By contrast, no clear international time trends are evident for HL incidence, either overall or for ages 15 to 44 or 45 to 85 years and greater [28]. Established risk factors for NHL analyzed as a single disease entity include older age, male sex, family history of hematopoietic malignancy, severe congenital or acquired immune suppression (e.g., ataxia telangiectasia, HIV or AIDS, or organ transplantation), and certain autoimmune disorders (e.g., Sjögren syndrome) [31e34]. In addition to HIV, which increases the risk of numerous NHL subtypes [35,36], other specific infectious agents are also etiologically linked to certain NHL subtypes. Causal relationships have been established between Epstein-Barr virus (EBV) and immunodeficiency-associated NHL, Burkitt lymphoma, and extranodal nasal natural killer/T-cell lymphoma; between human herpesvirus 8 (Kaposi sarcoma herpesvirus) and primary effusion lymphoma; between human T-cell lymphotropic virus 1 and adult T-cell leukemia/lymphoma; between Helicobacter pylori and gastric

mucosaeassociated lymphoid tissue lymphoma; and between hepatitis C virus and certain B-cell NHL subtypes [37e40]. Other relationships that are not established as causal include positive associations with type 2 diabetes mellitus [41,42], trichloroethylene exposure [43e45], and certain genetic polymorphisms in the tumor necrosis factor, lymphotoxin-alpha, interleukin 10 genes [46,47], and an inverse association with sun exposure [48,49]. In addition, higher body mass index is associated with increased risk of diffuse large B-cell lymphoma [50], cigarette smoking is associated with increased risk of follicular lymphoma [51], and alcohol consumption is associated with lower risk of Burkitt lymphoma [52]. Genomewide association studies have identified 16 genetic variants associated with risk of common NHL subtypes [53e59]. Other potential risk factors, such as other infectious agents [37,38,60], specific pesticides [2,61], diet [62], and reproductive and hormonal factors [63,64], are not consistently associated with risk of NHL or specific NHL subtypes. HL incidence varies bimodally by age, with an initial peak in young adults under 40 years and a second peak after around the age of 50 years [65,66]. HL incidence patterns differ by age group, with young-adult HL occurring more often in whites and slightly more often in women but older-adult HL occurring more often in nonwhites and men [67,68]. These patterns are reflected in histologic subtype-specific patterns. Nodular sclerosis HL, the most common subtype, affects mainly younger adults, whereas mixed-cellularity HL, the second most common subtype, occurs more frequently in older adults. Some established risk factors for HL are shared with NHL, including a family history of hematopoietic malignancy, severe congenital or acquired immune suppression, and certain autoimmune disorders [68e71]. EBV is an established cause of HL [40], with a higher proportion of EBV-positive (and mixed cellularity) tumors among older adults, young-adult males, nonwhites, and less economically developed populations [72]. Young-adult HL risk is consistently associated with indicators of a childhood social environment that favors delayed infection with common childhood infectious agents [68e71]. Delayed EBV infection, which manifests as infectious mononucleosis, increases the risk of tumor EBVe positive HL by approximately four-fold in young adults [73,74]. Certain genetic variants in the human leukocyte antigen (HLA) region, which encodes key molecules involved in immune function, are also consistently associated with HL risk [75,76]. The importance of the HLA region in HL etiology has been confirmed by four genomewide association studies [77e80], which also identified other putative susceptibility loci in other regions and demonstrated evidence of heterogeneity by tumor EBV status [80]. Cigarette smoking is positively associated with risk of EBV-positive and mixed-cellularity HL [81], and obesity appears to increase HL risk [50], whereas alcohol consumption may be inversely associated [82] and sun exposure may decrease EBV-positive HL risk [83]. Otherwise, few HL risk factors have been confirmed across multiple studies [67e71]. MM incidence increases steadily with age after approximately 40 years [25]. Rates are higher in males than females and, in the United States, in blacks than whites, who in turn have higher rates than Asians and Pacific Islanders [28]. Besides these demographic characteristics and a family history of MM [84,85], and perhaps obesity, which appears to be associated with an approximately 20% increase in MM risk [86], virtually no risk factors for MM are well established [87e89]. Profound immune suppression in the context of HIV infection or organ transplantation has been shown in several studies to increase MM risk, albeit to a lesser extent than risk of NHL and HL [90]. Certain autoimmune diseases and infections other than HIV have been shown in some but not all studies to be linked to higher risk of

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

MM [89,91e93]. Alcohol consumption [94], tobacco smoking [95], ionizing radiation [96], and other potential risk factors [87e89] are not consistently or convincingly associated with MM risk. Two genomewide association studies of MM have identified seven potential susceptibility loci that are robustly associated with MM [97,98]; these findings await follow-up functional studies. In general, epidemiologic studies of lymphoid malignancies are complicated by the heterogeneity of these disease entities. NHL comprises more than 40 histologic subtypes, some of which can be further subclassified by age group, anatomic site, gene expression pattern, or other features [1]. HL includes five histologic subtypes and also exhibits distinct risk factor profiles by age group and tumor EBV status [67]. Genetic and molecular subtypes of MMdwhich has traditionally been treated as distinct from the lymphomas, but is now classified with other NHL subtypes as a mature B-cell neoplasm [1]dare also recognized [99], but whether these subtypes have different risk factors is unknown. Etiologic variation among histologic subtypes of lymphoid malignancies may be obscured or distorted by combining all NHL, HL, or MM subtypes for analysis. In addition, various studies may use different disease classification systems or have dissimilar distributions of subtypes among cases. Thus, the inherent diversity of the lymphoid malignancies may be partially responsible for the inconsistency of results across studies and the few established risk factors. Methods The specific research question to be addressed by this review is whether the published epidemiologic literature supports a causal association between exposure to Agent Orange or TCDD and risk of NHL, HL, or MM. To resolve this question, this review includes three sets of epidemiologic studies of NHL, HL, and/or MM incidence or mortality: (1) studies of Vietnam veterans who were involved in the spraying or other handling of Agent Orange or otherwise reported exposure to Agent Orange; (2) studies of workers involved in the production or use of herbicides potentially contaminated by TCDD or of communities exposed via industrial accidents to herbicides or intermediates potentially contaminated by TCDD; and (3) studies of hospitalized patients or general community members with biological measurements of TCDD exposure. To provide context for the interpretation of studies in set 1, studies of Vietnam veterans without information on Agent Orange or TCDD exposure are also discussed briefly in this review; however, these studies are not included in the evaluation of the weight of evidence regarding Agent Orange or TCDD and lymphoid malignancies. As in previous reviews of the epidemiology of Agent Orange or TCDD exposure and cancer [100,101], to improve the specificity of the included evidence, studies (other than those of Vietnam veterans) were excluded if they did not estimate associations with exposure to Agent Orange or TCDD in particular. Thus, we excluded studies of workers exposed to herbicides not contaminated by TCDD, agents contaminated by polychlorinated dibenzo-p-dioxins other than TCDD, or groups of pesticides or herbicides not uniformly contaminated by TCDD. We also excluded studies of occupational groups with potential but unmeasured exposure to TCDD, such as farmers, forestry workers, pulp and paper workers, chloralkali workers, waste incinerator workers, steel mill workers, and fishermen because the independent effect of TCDD was not estimated in these studies, and average TCDD exposure levels in these groups of workers are relatively low [102]. We included ecologic studies that estimated exposure to TCDD or that documented TCDD exposure in the target population, but excluded those that estimated exposure to dioxins or dioxin-like agents in general. The health outcomes of interest in this review are NHL, HL, and MM analyzed as distinct disease entities or combined as

277

lymphomas (NHL and HL) or lymphoid malignancies (NHL, HL, and MM). Studies that combined all lymphopoietic malignancies (including leukemias) for analysis were excluded. Analytic epidemiologic studies of all designs were included as long as they met the eligibility criteria for specificity of exposure and outcome definitions. Relevant studies published as of June 2014 were identified by searching the MEDLINE database using PubMed and Ovid, as well as the Scopus database, using the following search strings: (“Agent Orange” OR TCDD OR dioxin* OR chloracne OR 2,3,7,8tetrachlorodibenzodioxin OR tetrachlorodibenzodioxin OR dioxins [MeSH] OR tetrachlorodibenzodioxin [MeSH]) AND (lymphoid* OR lymphoma* OR lymphatic* OR lymphop* OR hematolog* OR hematopo* OR haemato* OR NHL OR Hodgkin* OR “Hodgkin’s disease” OR “Hodgkin disease” OR “Hodgkin’s lymphoma” OR “Hodgkin lymphoma” OR cancer* OR lymphoma [MeSH] OR myeloma OR plasmacytoma OR multiple myeloma [MeSH] OR neoplasms [MeSH]) (Vietnam OR “Viet Nam”) AND (military OR war OR veteran* OR servicemen) AND (lymphoid* OR lymphoma* OR lymphatic* OR lymphop* OR hematolog* OR hematopo* OR haemato* OR NHL OR Hodgkin* OR “Hodgkin’s disease” OR “Hodgkin disease” OR “Hodgkin’s lymphoma” OR “Hodgkin lymphoma” OR cancer* OR lymphoma [MeSH] OR myeloma OR plasmacytoma OR multiple myeloma [MeSH] OR neoplasms [MeSH] OR mortality [ti] OR morbidity [ti]). Additional relevant studies were identified from reference lists of reviewed articles, the International Agency for Research on Cancer (IARC) monographs on TCDD [20,102], and the IOM reports [10e19]. When several reports were available from the same study, we considered only the most recent or complete results. Data items extracted from each eligible study were study design, characteristics of study subjects, period of exposure and follow-up (if any), exposure assessment methods, outcome assessment methods, number of cases in each exposure category, and relative risk (RR) point and interval estimates. The characteristics and results of all included studies are summarized in Supplementary Tables 1e7 as follows: Supplementary Table 1, design of all included epidemiologic studies of Agent Orange or TCDD exposure; Supplementary Table 2, results of studies of Vietnam veterans with information on exposure to Agent Orange or TCDD; Supplementary Table 3, results of studies of manufacturers and sprayers of herbicides potentially contaminated with TCDD; Supplementary Table 4, results of studies of the Seveso, Italy, industrial accident involving TCDD; Supplementary Table 5, results of studies of hospitalized patients or community members with biological measurements of TCDD exposure; Supplementary Table 6, design and results of studies of Vietnam veterans versus other groups without information on exposure to Agent Orange or TCDD; and Supplementary Table 7, design and results of ecologic studies of residential proximity to a TCDD source. All relevant results are provided in the tables; for readability, only selected illustrative results are provided in the text of this article. To avoid selecting a single, potentially nonrepresentative RR estimate from each study, obscuring information on exposureresponse trends, inappropriately equating disparate exposures such as varying levels of serum TCDD and self-reported Agent Orange exposure, and overstating the homogeneity and statistical precision of results [103e105], we did not conduct meta-analyses to combine the results of epidemiologic studies. Individual studies were evaluated in terms of the strengths and limitations of their study design, exposure classification, outcome classification, potential for confounding, potential for bias, and statistical approach. On the basis of all included studies, the weight of epidemiologic evidence to support a causal relationship between

278

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Agent Orange or TCDD and NHL, HL, or MM was assessed according to the Bradford Hill guidelines, namely, strength of association, consistency, specificity, temporality, biological gradient, plausibility, coherence, experiment, and analogy [106]. These guidelines were used not as strict criteria that must be fulfilled to establish causation, but rather as a sound and logical framework by which to synthesize the available epidemiologic evidence for or against a causal conclusion. Studies of Vietnam veterans without estimated Agent Orange or TCDD exposure and ecologic studies were given minimal consideration in the weight-of-evidence evaluation because these studies have major methodologic limitations that prevent causal inference and because their collective results do not contradict those of other reviewed studies. Results Studies of veterans with estimated Agent Orange or TCDD exposure The seminal study of the health effects of Agent Orange exposure is the Air Force Health Study, a prospective matched cohort study of 1261 male veterans of Operation Ranch Hand, the U.S. Air Force program that conducted aerial spraying of Agent Orange and other herbicides in South Vietnam from 1962 through 1971 [107]. Ranch Hand veterans were compared with up to 19,101 unexposed male U.S. Air Force veterans assigned to aerial cargo missions in Southeast Asia during the same time period. Both groups were followed for mortality starting from the first qualifying tour of duty onward based on record linkages to death certificates. Morbidity follow-up (in a subset of comparison subjects individually matched to Ranch Hand veterans on date of birth, race, rank, and military occupation) was conducted based on periodic physical examinations and questionnaires from 1982 through 2002, with selfreported and unreported disease diagnoses confirmed through medical records review. Average serum TCDD levels were markedly higher in Ranch Hand veterans than comparison subjects in 1987 (Ranch Hand veterans: median ¼ 12.4 ppt; range ¼ 0.0e617.7 ppt; comparison subjects: median ¼ 4.2 ppt; range ¼ 0.0e54.8 ppt) [108] and in 2002 (Ranch Hand veterans: median ¼ 5.0 pg/g lipid; 95% confidence interval [CI] ¼ 4.6e5.5; comparison subjects: median ¼ 2.2 pg/g lipid, 95% CI ¼ 2.1e2.2), whereas levels of other dibenzo-pdioxin congeners, dibenzofurans, and polychlorinated biphenyls did not differ meaningfully between the groups [109]. Although several publications from the Air Force Health Study reported on numbers of NHL, HL, and/or MM cases and deaths (e.g., [108,110e113]), only the most recent results and those that enabled calculation of RR estimates are described here and listed in Supplementary Table 2. Through 1993, no Ranch Hand veterans had died from reticulosarcoma or reticulolymphosarcoma (which are now considered part of NHL), chronic lymphocytic leukemia (CLL, which is now classified as an NHL), or HL; one had died from another type of lymphoma (probably now included with NHL) and one had died from MM (Supplementary Table 2) [112,114]. Compared with the expected number of deaths from each of these lymphoid malignancies among comparison subjects, the standardized mortality ratios (SMRs) were 0.0 (95% CI: 0e20.6) for reticulosarcoma, 0.0 (0e40.1) for reticulolymphosarcoma, 0.0 (0e13.8) for CLL, 0.0 (0e26.4) for HL, 2.2 (0.06e12.3) for other lymphoma, and 1.2 (0.03e6.4) for MM. (Throughout this article, RR point estimates are followed by 95% CIs unless otherwise stated; where CIs were not reported by the authors, Fisher exact 95% CIs were calculated.) With additional mortality follow-up through 1999, only one more MM death had occurred among Ranch Hand veterans, with no deaths from NHL, reticulolymphosarcoma, or HL (numbers of expected deaths not reported) [115].

In the Air Force Health Study morbidity study, in which 1197 Ranch Hand veterans and 1855 matched comparison subjects completed physical examinations and surveys from 1982 through 2002, two Ranch Hand veterans (one with 10 ppt serum TCDD back-extrapolated to the end of the qualifying tour of duty in Vietnam; one with >118.5 ppt TCDD) had developed NHL at the end of follow-up versus seven comparison subjects (five with 10 ppt TCDD, one with >10 ppt TCDD, one with unknown TCDD; Supplementary Table 2) [107]. Five Ranch Hand veterans developed CLL (three with 10 ppt TCDD, one with >10 to 118.5 ppt TCDD, one with unknown TCDD) versus six comparison subjects (all with 10 ppt TCDD). One Ranch Hand veteran (with 10 ppt TCDD) developed HL versus three comparison subjects (all with 10 ppt TCDD). Four Ranch Hand veterans developed MM (one with 10 ppt TCDD, three with >10 to 118.5 ppt TCDD) versus one comparison subject (with 10 ppt TCDD). RRs for Ranch Hand veterans versus comparison subjects or by serum TCDD level were not estimated. (Another analysis in the Air Force Health Study reported results for all lymphopoietic malignancies combined, including leukemia, but not for NHL, HL, and/or MM [116].) Substantial strengths of the Air Force Health Study include its focus on the group of Vietnam veterans most highly exposed to Agent Orange [108,109,117e119], its prospective long-term followup, enrollment of a highly comparable matched group of unexposed subjects to minimize confounding, validation of reported and unreported diagnoses based on medical records, and measurement of serum TCDD in 86% of Ranch Hand veterans and 82% of comparison subjects to enable direct identification of those who were most highly exposed. However, the numbers of lymphoid malignancy cases and deaths were small, and adjusted RRs were not reported for NHL, HL, or MM incidence. Therefore, although there appeared to be no significant excess of these cancers among Ranch Hand veterans, the study size was insufficient to rule out small differences between exposure groups. The Army Chemical Corps study is the other investigation of Vietnam veterans highly exposed to Agent Orange, namely, members of U.S. Army Chemical Corps units that were responsible for the storage, handling, mixing, application, and equipment maintenance of herbicides, including Agent Orange, and other chemicals in South Vietnam between 1966 and 1971 [120,121]. Mean lipidcorrected serum TCDD levels in Army Chemical Corps veterans were 4.3 ppt (range: 0.8e85.8 ppt) among those who were stationed in Vietnam and reported having sprayed herbicides, 3.1 ppt (range: 0.8e9.6 ppt) among those who were not stationed in Vietnam but sprayed herbicides, 2.70 ppt (range: 0.6e27.7 ppt) among those who were stationed in Vietnam but did not spray herbicides, and 2.1 ppt (range: 0.4e12.5 ppt) among those who were neither stationed in Vietnam nor sprayed herbicides [118]. Levels of six other dioxin congeners did not differ significantly between Vietnam sprayers and non-Vietnam nonsprayers [117]. Thus, serum TCDD levels were lower in Army Chemical Corps veterans who sprayed herbicides in Vietnam than in Ranch Hand veterans but higher than in nonsprayers and those who served outside Vietnam. Among 894 Army Chemical Corps veterans who were stationed in Vietnam, none developed or died from NHL (SMR ¼ 0.0 [0.0e9.2]; Supplementary Table 2) [120]. Two cases developed histologically confirmed HL, including one identified from the Veterans Administration (VA) Agent Orange Registry of Vietnam veterans who volunteered to undergo a special physical examination at a VA medical center towing to presumed Agent Orange exposure (standardized incidence ratio [SIR] within the Agent Orange Registry ¼ 5.0 [0.1e27.9]) and one identified from the VA Patient Treatment File of inpatient discharges from all VA medical centers in the U.S. (combined SIR ¼ 2.9 [0.3e10.3]).

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

In a subsequent expanded study, 2872 Army Chemical Corps veterans who served in Vietnam were compared with 2737 Army Chemical Corps veterans of the same era who did not serve in Southeast Asia [121]. Four deaths from lymphoid cancers (calculated as the difference between all lymphopoietic cancers and leukemia) occurred in the Vietnam cohort and two occurred in the non-Vietnam cohort, resulting in a crude RR of 1.93 (0.35e1.51; calculated based on reported person-time at risk). Compared with males in the general U.S. population, the SMR for lymphoid cancers was 0.48 (0.13e1.24) among Vietnam Army Chemical Corps veterans and 0.33 (0.04e1.20) among non-Vietnam Army Chemical Corps veterans. The strengths of the Army Chemical Corps study include its focus on a group of Vietnam veterans with relatively high exposure to Agent Orange and measurement of serum TCDD in a subset of subjects to confirm higher exposure among herbicide sprayers assigned to Vietnam [120,121]. The more recent study was further strengthened by its internal comparison to Army Chemical Corps veterans who served outside of Southeast Asia [121]. However, this study was limited by its small size and lack of detailed exposure information. Another epidemiologic study that estimated the association between Agent Orange exposure and risk of NHL and HL is the Selected Cancers Study, a case-control study of 1157 men with incident, pathologically confirmed NHL and 310 with HL diagnosed between 1984 and 1988 in eight U.S. regions covered by populationbased cancer registries, compared with 1776 controls identified by random-digit dialing [122,123]. Exposure information was collected by telephone interview. Military service in Vietnam was associated with a significant increase in NHL risk (odds ratio [OR] vs. Vietnam-era veterans who did not serve in Vietnam ¼ 1.52 [1.00e2.32]; 96 cases) but not in HL risk (OR ¼ 1.23 [0.65e2.36]; 26 cases; Supplementary Table 2) [122,123]. However, among Vietnam veterans, NHL risk was higher (albeit sometimes nonsignificantly so) among those expected to have had lower exposure to Agent Orange, including Navy veterans (OR vs. nonveterans ¼ 1.89 [1.11e3.24]; 32 cases), Marine Corps veterans (OR ¼ 1.84 [0.78e4.34]; 10 cases), officers (OR ¼ 1.78 [0.80e3.96; 12 cases), veterans of noncombat support units (OR ¼ 1.50 [1.02e2.21]; 56 cases), those who served in I Corps (OR vs. those who did not serve in Vietnam ¼ 2.25 [1.21e4.18]; 23 cases), and those who served in the blue-water Navy (OR ¼ 2.18 [1.23e3.88]; 28 cases). By contrast, NHL risk was not increased among veterans who served in III Corps, which was stationed in the region and time period of heaviest spraying (OR ¼ 0.89 [0.50e1.58]; 19 cases). HL risk also was not significantly increased among Vietnam veterans expected to have had greater opportunity for Agent Orange exposure. Participants in the Selected Cancers Study were asked to report self-perceived exposure to herbicides while in Vietnam, although such self-reported information has been shown not to accurately reflect serum TCDD levels among U.S. veterans [119]. None of the indicators of potential Agent Orange exposure were significantly associated with NHL or HL risk, including having passed through a defoliated area in Vietnam (NHL OR ¼ 0.82 [0.45e1.49]; HL OR ¼ 0.65 [0.25e1.67]; 27 NHL cases, 8 HL cases), having had any possible contact with Agent Orange (NHL OR ¼ 1.08 [0.58e2.02]; HL OR ¼ 0.76 [0.28e2.07]; 26 NHL cases, 7 HL cases), having been present when others sprayed Agent Orange (NHL OR ¼ 0.98 [0.39e2.48]; HL OR ¼ 0.74 [0.15e3.67]; 9 NHL cases, 2 HL cases), having gotten Agent Orange on skin or clothes (NHL OR ¼ 1.08 [0.40e2.96]; HL OR ¼ 0.47 [0.06e3.99]; 8 NHL cases, 1 HL case), having handled equipment or containers used with Agent Orange (NHL OR ¼ 0.41 [0.04e4.07]; 1 NHL case, 0 HL cases), or having

279

sprayed Agent Orange (0 NHL cases, 0 HL cases; Supplementary Table 2) [122,123]. The Selected Cancers Study has several strengths, including pathologic confirmation of cases, relatively high participation rates (88% among cases and 83% among controls within households that completed eligibility screening [90%] and identified an eligible man [91%]), and detailed information on military history and numerous potential confounders [122,123]. However, a key limitation is the lack of information on serum TCDD levels to validate self-reported Agent Orange exposure. Given the poor correlation between selfreported Agent Orange exposure and serum TCDD levels [119], the lack of a significant association between self-perceived opportunities for Agent Orange exposure and NHL or HL risk in this study does not necessarily rule out a true association. Moreover, serum TCDD levels have been shown not to differ significantly between Vietnam veterans who served in III Corps and non-Vietnam veterans (median ¼ 3.8 ppt in both groups in 1986e1987) nor do TCDD levels correspond to a greater likelihood of Agent Orange exposure based on military records [119]. Therefore, the lack of a significant elevation in NHL or HL risk among Vietnam veterans whose military service history conferred a higher probability of Agent Orange exposure does not provide strong evidence against a causal effect of Agent Orange or TCDD. However, this study also provides no evidence to support such an effect. The authors concluded that their “results strongly argue against the possibility that exposure to Agent Orange is responsible for the increased risk of NHL among Vietnam veterans” [123]. Two hospital-based case-control studies, one of NHL and the other of HL, were conducted among male Vietnam-era veterans born between 1937 and 1954 and treated at one of 82 U.S. VA hospitals with at least 10 lymphoma cases between 1969 (or 1975, for HL) and 1985 [124,125]. Cases were 201 NHL patients and 283 HL patients with pathologically confirmed disease identified from the VA Patient Treatment File. Controls with another diagnosis and no lymphoma history were matched 2:1 to cases on hospital, year of first VA hospital discharge, and birth year. Military service in Vietnam between 1965 and 1973 was ascertained from the National Personnel Records Center. Indirect indicators of potential Agent Orange exposure were service in the Army or Marine Corps, combat duty, service between 1967 and 1969, service in III Corps, and longer duration of service in Vietnam. None of the proxies for Agent Orange exposure were significantly associated with risk of NHL or HL (Supplementary Table 2) [124,125]. For example, the OR for NHL in association with Army service in III Corps, compared with Army service exclusively outside of Vietnam, was 0.70 (0.40e1.23; 26 exposed cases) and that for HL, compared with never having served in Vietnam, was 1.25 (0.78e2.00; 40 exposed cases). The ORs for NHL in association with having served on combat duty in the Army or Marine Corps, compared with service in the same branch outside of Vietnam, were 0.91 (0.50e1.65) and 0.89 (0.25e3.08), respectively, whereas the OR for HL in association with combat duty in the Army or Marine Corps, compared with never having served in Vietnam, was 1.28 (0.81e2.03). Strengths of these studies include the pathologic confirmation of cases, the low potential for selection bias because of the lack of active subject contact, and the use of objectively recorded and obtained data on military service history [124,125]. However, because subjects were not contacted, detailed exposure data were unavailable to estimate associations with exposure to Agent Orange or any other exposures particular to military service in Vietnam. As stated earlier, serum TCDD levels are poorly correlated with estimated exposure to Agent Orange based on military service records [119]. Consequently, although the results of this study do not

280

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

support a causal effect of Agent Orange or TCDD on NHL or HL, they also cannot rule out such an effect. The most recent group in which associations have been estimated between Agent Orange exposure and risk of NHL, HL, and MM is the cohort of South Korean veterans of the Vietnam War [126,127]. Studies in this group include a cross-sectional analysis of self-reported cancer prevalence based on a postal survey completed by 114,562 veterans [126] and a retrospective cohort study of cancer incidence among 180,251 veterans linked to the national cancer incidence database [127]. Potential exposure to Agent Orange was evaluated based on the proximity of the respondent’s division or brigade (and, in the cross-sectional study, based on the proximity of the respondent’s self-reported battalion or company) to sprayed areas, using the Agent Orange Exposure Opportunity Index model E4 developed by Stellman et al. [128]. Agent Orange exposure was alternatively classified based on six survey questions in the crosssectional study. In the cross-sectional analysis [126], self-perceived Agent Orange exposure was significantly associated with the prevalence of all three of the lymphoid malignancies, including NHL (OR for high vs. no exposure ¼ 2.93 [1.47e5.82; Ptrend across four categories < .001; 59 cases with high exposure), HL (OR ¼ 4.01 [1.92e8.36]; Ptrend < .001; 45 cases), and MM (OR ¼ 3.69 [3.02e4.52]; Ptrend < .001; 607 cases; Supplementary Table 2). By contrast, division- or brigade-level modeled exposure was not significantly associated with risk of any of the three outcomes (NHL OR for high vs. no exposure ¼ 0.97 [0.44e2.13]; Ptrend ¼ .66; HL OR ¼ 0.68 [0.27e1.68]; Ptrend ¼ .63; MM OR ¼ 1.14 [0.90e1.45]; Ptrend ¼ .18; 37 NHL cases, 30 HL cases, and 308 MM cases with high exposure). Battalion- or company-level modeled exposure also was not significantly associated with risk of NHL (OR ¼ 0.95 [0.49e1.87]; Ptrend ¼ .64; 28 cases) or HL (OR ¼ 0.76 [0.32e1.81]; Ptrend ¼ .38; 19 cases), but it was marginally significantly associated with increased risk of MM (OR ¼ 1.25 [0.99e1.56]; Ptrend ¼ .048; 322 cases). In the retrospective analysis of cancer incidence [127], proximity-based estimated Agent Orange exposure was not significantly associated with risk of NHL (hazard ratio for high vs. low exposure ¼ 1.09 [0.81e1.47]; 96 cases with high exposure), HL (hazard ratio ¼ 1.27 [0.41e3.93]; 7 cases), or MM (hazard ratio ¼ 1.14 [0.65e2.01]; 28 cases; Supplementary Table 2). Additional analyses using estimated exposure categorized into four groups, classifying exposure as a continuous variable, restricting the analysis to veterans who served in units with a defined tactical area, restricting to those who served for more than 6 months, or restricting to enlisted men did not reveal evidence of a significant positive association. Although statistically strengthened by large numbers [126], the cross-sectional analysis is limited by unreliable outcome data. Self-reported cancer diagnoses were not validated based on medical records, and the fact that the frequency of self-reported MM was roughly 10 times that of self-reported NHL in this study suggests that outcome data were highly misclassified, given that the true incidence and prevalence rates of NHL in South Korean males aged 50 to 69 years are approximately triple those for MM [25]. (One possible interpretation of the results for MM is that they were for malignant melanoma; however, this possibility cannot be verified.) Moreover, the overall prevalence rates of selfreported NHL, HL, and MM among study subjects were 71, 58, and 716 per 100,000, respectively, whereas the corresponding fiveyear prevalence rates estimated by IARC for males of all ages in the Republic of Korea in 2012 are 34.4, 2.9, and 6.5 per 100,000 [25]. These discrepancies again suggest substantial misclassification of prevalent cancer. Both the studies are also undermined by untrustworthy exposure data [126,127]. Among 102 South Korean veterans of the

Vietnam War in 2007, serum TCDD levels were not significantly correlated with self-reported perceived exposure to Agent Orange (r2 ¼ 0.129), division- or brigade-level proximity-based exposure (r2 ¼ 0.073), or battalion- or company-level proximity-based exposure (r2 ¼ 0.159) [129]. Moreover, average serum TCDD levels (mean ¼ 1.2 ppt, median ¼ 0.9 ppt) were lower than in U.S. veterans stationed in the continental United States or Germany during the Vietnam War (mean ¼ 4.1 ppt) [119] and comparable to those in the U.S. general population (mean z 1.1 ppt) [130]. The accuracy and precision of the Exposure Opportunity Index model used in this study have also been called into question [131,132], and the IOM Committee cannot independently support the validity of any proposed exposure model [10]. Thus, none of the exposure metrics used in these studies can be interpreted as a valid proxy for Agent Orange exposure. In particular, self-reported exposure was almost certainly differentially biased in favor of a positive association with self-reported disease prevalence, as reflected in the uniformly statistically significant positive trends detected (Ptrend < .001 for 54 of 54 disease end points tested) [126]. Overall, these studies provide no reliable information to support or refute an association between Agent Orange or TCDD exposure and risk of NHL, HL, or MM. In summary, none of the studies of Agent Orange or TCDD exposure and risk of NHL, HL, or MM among Vietnam veterans, including the Air Force Health Study [107,112,114,115], the Army Chemical Corps Study [120,121], the Selected Cancers Study [122,123], the VA hospitals study [124,125], and the South Korean veterans studies [126,127], offer credible evidence to support a causal association. The substantial limitations of these studies, including small size of the first two studies and the lack of valid estimates of Agent Orange exposure in the others, as well as probably invalid outcome information in the Korean cross-sectional study, prevent them from ruling out an association. Nevertheless, the available epidemiologic evidence does not demonstrate any significant association between valid Agent Orange or TCDD exposure and risk of NHL, HL, or MM among Vietnam veterans. Studies of Vietnam veterans without estimated Agent Orange or TCDD exposure The significant positive association between Vietnam military service and risk of NHL (but not HL) in the Selected Cancers Study [122,123] raises the question of whether this result is explained by causal exposures, bias, or chance. To evaluate the consistency of the association between military service in Vietnam and risk of lymphoid malignancies, studies of Vietnam veterans compared with other groups, in the absence of information on potential Agent Orange or TCDD exposure, are addressed in this section and summarized in Supplementary Table 6. Differences in health outcomes between Vietnam veterans and veterans who served outside of Southeast Asia or general populations cannot reasonably be attributed to Agent Orange exposure or any other unmeasured exposure related to Vietnam, Southeast Asia, or military service. For example, hepatitis B virus is endemic in Vietnam and Southeast Asia [133], and the prevalence of hepatitis C virus is higher among veterans, especially Vietnam veterans, than in the general U.S. population [134e136]. Evidence strongly implicates hepatitis C virus as a cause of NHL, and hepatitis B virus appears to be positively associated with risk [40]. HIV infection is also more common among U.S. veterans [137] than in the general U.S. population [138] and increases the risk of NHL, HL, and possibly MM [90]. Additional unidentified exposures other than Agent Orange, none of which were controlled for in these studies, may also be responsible for observed differences in risk of lymphoid malignancies between Vietnam veterans and other groups. Because such studies

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

provide essentially no information on Agent Orange or TCDD exposure, they are discussed only briefly here. No significant difference in NHL incidence or mortality was reported in the majority of available studies (Supplementary Table 6). These included investigations that compared male U.S. military veterans stationed in Vietnam versus non-Vietnam veterans seen at any of 82 VA medical centers across the United States [124] (described in the previous section); male Australian National Service conscripts who served in Vietnam versus those who served only in Australia, or versus the general male Australian population [139]; female Australian Vietnam veterans versus the general female population of Australia [140]; male U.S. Army veterans who served in Vietnam versus those who served elsewhere during the same era, or versus the general U.S. male population [141,142]; female veterans of the U.S. military who served in Vietnam versus those who served outside the Pacific theater, or versus the general U.S. female population [143]; U.S. Marine Corps Vietnam veterans versus those who served outside Vietnam or versus the general U.S. male population [144]; male New Zealand veterans with service in Vietnam versus the general New Zealand male population [145]; or male Korean veterans who served in Vietnam versus the general Korean male population (including no significant excess of overall, follicular, diffuse, diffuse large B-cell, T-cell, or peripheral T-cell NHL) [146]. A few exceptions to the mostly statistically nonsignificant findings were reported, but even these were not entirely consistent with increased NHL risk among Vietnam veterans. Self-reported NHL prevalence was significantly higher among male Australian Vietnam veterans than expected in the general population [147], but this difference was substantially attenuated (albeit still marginally significant) after accounting for invalid outcome information (Supplementary Table 6) [148]. However, this result was contradicted when validated diagnoses were ascertained by linkage to national registries, which showed significantly lower NHL incidence and mortality in Australian Vietnam veterans (including Army veterans) than expected based on national rates [149,150]. Mortality from lymphoid tissue cancers other than lymphosarcoma, reticulosarcoma, or HL was significantly higher among Wisconsin Vietnam veterans compared with Wisconsin nonVietnam veterans but not compared with the general Wisconsin or U.S. population (Supplementary Table 2) [151]. RRs for mortality from lymphosarcoma and reticulosarcoma among Vietnam veterans were not estimated in this study based on four deaths among 43,398 Vietnam veterans and eight deaths among 78,940 nonVietnam veterans from the same era. However, based on calculated differences in observed and expected deaths between all lymphopoietic cancers and leukemia or aleukemia, mortality from lymphoid cancers overall was not significantly higher than expected among Wisconsin Vietnam veterans compared with any referent group. Finally, the incidence of CLL was significantly higher among male New Zealand veterans who served in Vietnam than in the general male population of New Zealand (Supplementary Table 6) [145], but it did not differ significantly between male Australian National Service conscripts who did and did not serve in Vietnam [139]. Of note, most Australian and New Zealand military personnel who served in Vietnam were stationed in Phuoc Tuy Province, which was not heavily sprayed with Agent Orange [145,152]. For HL, results were also mostly statistically null (Supplementary Table 6). No significant excess of incidence or mortality was found among male veterans of the U.S. military who were assigned to Vietnam compared with non-Vietnam veterans seen at any of 82 VA medical centers across the United States [125] (described in the previous section); male Australian National Service conscripts who served in Vietnam compared with those who

281

served only in Australia, or with the general male Australian population [139]; male U.S. Army veterans who served in Vietnam compared with those who did not [142]; male New Zealand veterans who served in Vietnam compared with the general New Zealand male population [145]; or male Korean veterans who served in Vietnam compared with the general Korean male population [146]. HL incidence was significantly higher in male Australian Vietnam veterans (including Army veterans) than expected in the general population of Australia [149] but HL mortality was not [150]. No significant excesses of MM were detected among studies of Vietnam veterans compared with other groups (Supplementary Table 6). Specifically, MM incidence and mortality did not differ significantly between male Australian National Service conscripts who did and did not serve in Vietnam [139]; between male U.S. Army Vietnam veterans and non-Vietnam veterans [142]; between male New Zealand Vietnam veterans and the general New Zealand male population [145]; or between male Korean Vietnam veterans and the general Korean male population [146]. MM incidence was significantly lower in male Australian Vietnam veterans (including Army veterans) than expected in the general Australian population [149] but MM mortality was not [150]. Proportionate mortality and morbidity studies, which cannot distinguish whether exposure increases the risk of death from the cause of interest or prevents death from other causes [153], and which also usually lack detailed exposure information, are mentioned here in passing only for the sake of completeness. Among proportionate mortality and morbidity studies of U.S. Vietnam veterans in New York State [154], Wisconsin [151], West Virginia [155], Massachusetts [156e158], Michigan [159], and the U.S. Army and Marine Corps [160,161], statistically significant excesses were observed for HL mortality in West Virginia Vietnam veterans versus non-Vietnam veterans [155], NHL mortality in Michigan Vietnam veterans versus non-Vietnam veterans [159], and NHL and HL mortality in U.S. Marine Corps (but not Army, including those who served in I Corps [161]) Vietnam versus nonVietnam veterans [160]. Otherwise, differences for Vietnam veterans were statistically nonsignificant. Taken together, these studies show no consistent association between Vietnam military service and NHL, HL, or MM incidence or mortality. Most estimated associations within and across studies were statistically nonsignificant. Therefore, the significant positive association with Vietnam military service (but not potential Agent Orange exposure) observed in the Selected Cancers Study [123] was probably because of chance, confounding, or bias. That association was detected both with minimal control and with extensive control for confounding, suggesting that the minimal control for confounders in most studies described in this section was not responsible for the failure to detect a significant association. In the absence of information on specific exposures, these studies collectively provide essentially no insight regarding potential risk associations with Agent Orange or TCDD exposure in Vietnam. Studies of manufacturers and sprayers of herbicides potentially contaminated with TCDD The industry-based studies that are most specific to occupational exposure to TCDD are those that concern chemical workers involved in the production of potentially TCDD-contaminated herbicides (2,4,5-trichlorophenoxyacetic acid [2,4,5-T], trichlorophenol, and/or pentachlorophenol) and sprayers of the same herbicides (mainly 2,4,5-T). A review of the cancer results from these studies is complicated by the partial overlap of the study subjects included in various reports. IARC [162] coordinated a pooled analysis of several studies (most of which had been reported

282

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

separately) from Europe and Oceania. The National Institute for Occupational Health and Safety (NIOSH) [163] combined several U.S.-based studies (also reported separately) in a pooled analysis The results of the two pooled analyses were then reported in a combined publication, which included four additional German cohorts that also had been reported separately [164]. The NIOSH pooled study was subsequently updated [165], as were analyses of some of the individual plants included in the NIOSH study, the IARC pooled study, and one of the German cohorts. We reviewed the combined publication of the two pooled analyses [164], the NIOSH study update [165] and the subsequent individual cohort studies that reported results for expanded study groups and/or longer follow-up [166e173]. An additional study of herbicide production workers was not included in any of the pooled studies and is discussed separately in our review [174]. The results of the relevant studies are summarized in Supplementary Table 3. The combined IARC/NIOSH pooled analysis comprised 21,863 male and female workers exposed to phenoxy herbicides or chlorophenols from 32 cohorts of manufacturing workers from Austria, Denmark, Finland, Germany, Italy, the Netherlands, New Zealand, Sweden, the United Kingdom, and the United States, and four cohorts of sprayers from Australia, Canada, New Zealand, and the United Kingdom [164]. Overall, the periods of employment and mortality follow-up were between 1939 and 1992, but they varied somewhat among the countries. Exposure to TCDD was defined based on 1) employment during manufacture or spraying of 2,4,5-T, pentachlorophenol, or other potentially contaminated compounds, or 2) employment in plants with documented exposure through environmental measurements or biomonitoring (data available for 573 workers from 10 cohorts). Substantially higher mean serum TCDD levels than in the population at large were found in subsets of workers, including phenoxy herbicide and chlorophenol production workers (means ¼ 17e233 ppt), especially those who developed chloracne (means ¼ 389e402 ppt), and sprayers (mean ¼ 53 ppt) [163,175e179]. From 28 pooled cohorts, 13,831 workers (63.3%) were classified as exposed to TCDD (73% manufacturing workers and 27% sprayers; 63% from the IARC study and 37% from the NIOSH study), and 7,553 workers (34.5%) were classified as unexposed to TCDD (479 workers from a UK plant [2.2%] could not be classified as to TCDD exposure). The SMR for NHL was 1.39 (0.89e2.06; 24 deaths) among exposed workers and 1.00 (0.46e1.90; 9 deaths) among unexposed workers (Supplementary Table 3) [164]. Among exposed workers, SMRs for NHL according to time since first exposure were 0.63 (0.08e2.26; 2 deaths) for 0-9 years, 1.45 (0.63e2.87; 8 deaths) for 10-19 years, and 1.63 (0.89e2.73; 14 deaths) for 20 or more years. The analysis by duration of exposure revealed no apparent trend; the SMR was higher in the group with less than one year of exposure (2.43 [1.21e4.35]; 11 deaths) than in the groups with longer exposure (Supplementary Table 3) [164]. In the analysis by job category, the SMR was higher in the group of workers other than primary production and maintenance (4.85 [2.33e8.93]; 10 deaths) than in other groups. The SMRs for HL were 1.29 (0.56e2.53; 8 deaths) among TCDD-exposed workers and 0.27 (0.01e1.51; 1 death) among the unexposed. Corresponding SMRs for MM were 1.21 (0.55e2.29; 9 deaths) and 1.59 (0.69e3.13; 8 deaths). Results of detailed analyses were not reported for HL and MM. A nested case-control study of NHL was conducted within the IARC cohort [180]. Cases comprised 20 workers who died from NHL, as indicated on the death certificate, and an additional 12 cases identified through linkage with cancer registries; histological verification was available for 30 of the 32 cases or deaths. A total of 158 controls were selected among workers in the cohort, who were free from NHL at the time of diagnosis of the cases, matched by sex,

age, and country. Exposure to TCDD and other 20 agents and mixtures was assessed for each subject by a team of industrial hygienists blinded to case-control status, based on job histories, company questionnaires, and other reports. Level of exposure was assessed on a relative scale, and cumulative exposure was calculated by multiplying level by duration of exposure. A five-year lag was applied; workers with cumulative exposure below 0.005 unityears (corresponding to approximately one day of exposure) were considered unexposed. A total of 11 cases and 39 controls were classified as ever exposed to TCDD (OR ¼ 1.93 [0.74e5.07]); ORs for cumulative exposure were 1.42 (0.44e4.64; 4 exposed cases) for low exposure (<1 unit-year), 3.63 (0.71e18.67; 3 exposed cases) for medium exposure (1 to < 10 unit-years), and 3.56 (0.66e19.20; 4 exposed cases) for high exposure (10 or more unit-years) (Ptrend ¼ .1) (Supplementary Table 3). With updated mortality follow-up through 1993 instead of 1987, as reported in the combined IARC/NIOSH pooled analysis [164], the NIOSH pooled study included 5132 workers employed in 12 plants, all of whom were considered exposed to TCDD [165]. The SMR for NHL, based on 12 deaths, was 1.10 (0.56e1.91); the SMR for HL, based on three deaths, was 1.09 (0.22e3.19) and that for MM, based on 10 deaths, was 2.07 (0.99e3.80) (Supplementary Table 3). Workers employed in one of the U.S. plants included in the NIOSH pooled analysis were studied by Collins et al. in two subsequent analyses [168,169]. In the first analysis [169] of 773 male pentachlorophenol manufacturing workers exposed to TCDD, the period of follow-up was not specified. Serum TCDD levels were available for 128 workers and were used to model TCDD exposure for the whole study cohort. The SMR for NHL in the whole cohort was 2.4 (1.0e4.7; 8 deaths; Supplementary Table 3). Exclusion of workers exposed to trichlorophenol resulted in a SMR of 2.8 (1.1e5.7; 7 deaths). When workers were categorized in tertiles according to estimated TCDD exposure, SMRs were 1.6 (0.2e5.7; 2 deaths) for low exposure (<0.15 ppb-years), 2.8 (0.6e8.1; 3 deaths) for medium exposure (0.15e0.824 ppb-years), and 3.1 (0.6e9.1; 3 deaths) for high exposure (0.825 ppb-years). The estimated RR of NHL per ppb-year toxic equivalency quotient (TEQ) was 1.006 (0.960e1.054). There were no deaths from HL in this cohort (expected deaths not reported) and two deaths from lymphohematopoietic cancers other than NHL, HL, and leukemia, a category that included MM and unspecified lymphomas (SMR ¼ 1.3 [0.2e4.6]). The second analysis [168], which extended mortality follow-up through 2003, was based on 1615 trichlorophenol workers with potential TCDD exposure, including 196 pentachlorophenol workers who presumably were also included in the first analysis. TCDD exposure was modeled based on the serum levels for 280 workers. After excluding pentachlorophenol workers, there were eight deaths from NHL (SMR ¼ 1.3 [0.6e2.6]; Supplementary Table 3). The RR for NHL death per ppb-year increase in cumulative estimated TCDD level was 1.011 (0.961e1.064). The SMR for HL, based on two deaths, was 2.0 (0.2e2.3) and that for “other lymphohematopoietic cancers” (see first analysis) was 0.7 (0.1e2.6; 2 deaths). A further analysis [172] included 2122 workers at four pentachlorophenol production plants that were part of the NIOSH pooled study [165] and partially overlapped with the cohorts studied by Collins et al. [168,169]. A subcohort of 720 workers involved in trichlorophenol production and considered exposed to TCDD was followed through 2005. Eight deaths from NHL occurred among the exposed workers (SMR ¼ 2.50 [1.08e4.93]), and nine occurred among the unexposed workers (SMR ¼ 1.41 [0.64e2.67]; Supplementary Table 3). Overall, there was no positive trend in NHL mortality with duration of employment. No deaths from HL were observed among exposed workers, whereas one such death occurred in the unexposed subcohort (SMR ¼ 0.97 [0.02e5.41]).

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

SMRs for MM were 0.72 (0.02e3.99; 1 death) among exposed workers and 1.84 (68e4.00; 6 deaths) among the unexposed. Two cohorts from New Zealand included in the IARC pooled analysis were followed for mortality through 2000 [173]. Subjects were 1025 production workers and 703 sprayers, including 813 and 699, respectively, classified as exposed. The SMRs for NHL were 0.87 (0.02e4.87; one death) among production workers and 0.69 (0.02e3.84; one death) among sprayers (Supplementary Table 3). Production workers also experienced one HL death (SMR ¼ 5.58 [0.14e31.0]) and three deaths from MM (SMR ¼ 5.51 [1.14e16.1]). No deaths from these neoplasms occurred among sprayers. The cohort study of New Zealand manufacturing workers was expanded in a subsequent analysis restricted to 1599 workers who were employed between 1969 and 1988, when 2,4,5-T was used and exposure to TCDD was possible, with mortality follow-up through 2004 [171]. A total of 1134 workers were classified as exposed to TCDD, whereas 465 workers were considered unexposed. Cumulative TCDD levels were estimated from serum of 346 workers. Four deaths from NHL were observed in this cohort; three deaths were among exposed workers (SMR ¼ 1.6 [0.3e4.7]) and one death was among unexposed workers (SMR ¼ 1.6 [0.0e8.7]; Supplementary Table 3). Results according to estimated cumulative TCDD exposure did not suggest a positive trend for NHL mortality. There was one death from HL (SMR ¼ 4.2 [0.1e23.3]) and there were two deaths from MM (SMR ¼ 2.2 [0.2e8.1]) among exposed workers. These deaths were likely to be the same as those included in the other analyses. No deaths from HL or MM were observed among unexposed workers. For 2106 male production workers employed in two Dutch plants included in the IARC pooled study, mortality follow-up was updated through 2006 [166]. On the basis of either their history of employment in selected departments (or having entered them on a regular basis) or exposure to an accident that occurred in 1963 in factory A, 539 workers in factory A and 411 workers in factory B were classified as exposed to TCDD. The remaining 482 workers in factory A and 626 workers in factory B were classified as unexposed. The RR of NHL for TCDD exposure was 0.92 (0.19e4.47) in factory A, based on four deaths among the exposed and three deaths among the unexposed. In factory B there was one NHL death among the exposed, and there were no deaths among the unexposed (Supplementary Table 3). One of the four exposed deaths in factory A was of an individual who was present during the 1963 accident. Boers et al. [167] extended this analysis by including estimated TCDD exposure based on plasma measurements available for 187 workers. The RR of NHL per log unit of TCDD (with a 10-year lag) was 1.36 (1.06e1.74; Supplementary Table 3). In this analysis, all workers in factory B were considered unexposed to TCDD, based on their geometric mean plasma TCDD level of 0.4 ppt. When the analysis was restricted to factory A, the RR per lagged log unit of TCDD level was slightly attenuated (RR ¼ 1.27 [0.95e1.71]). Results on HL and MM mortality were not presented in the two reports on the update of the Dutch cohort. In one of the plants from Germany included in the IARC pooled study [170], mortality follow-up was extended to 2007 for 1489 workers ever employed between 1952 and 1984, after which the plant was closed. A study pathologist coded cause of death for deceased cohort members. The expected numbers of deaths were based on regional rates. A total of seven deaths from NHL were observed (SMR ¼ 1.59 [0.64e3.28]; Supplementary Table 3). Results on HL and MM mortality were not reported. Serum levels from a subset of workers were used to estimate cumulative TCDD exposure, but results on NHL were not reported according to the estimated TCDD levels. The only group of TCDD-exposed workers not included in the IARC or NIOSH pooled analysis comprised 243 male workers who

283

participated between 1953 and 1969 in clean-up, repair, maintenance, or demolition activities at a German plant after a reactor accident in 1953 that led to TCDD release [174]. Serum TCDD levels were measured for all cohort members and were used to estimate TCDD doses of 0.1 mg/kg body weight for 135 workers (55.6%) and 1.0 mg/kg body weight for 69 (28.4%). Follow-up for mortality and cancer incidence extended from 1953 through 1992. Most analyses in this cohort were conducted for lymphatic and hematopoietic cancers combined, including leukemia. However, in the abstract the authors reported that no cases of NHL had been identified to date; thus, the SMR and SIR for NHL were both 0. Subsets of workers who developed chloracne, suggesting high TCDD exposure, were identified in the NIOSH pooled cohort (608 workers) [165] and in the German accident cohort (113 workers) [174]. Results for specific lymphoid neoplasms were not reported for these workers, but the SMR for lymphatic and hematopoietic malignancies combined in the NIOSH cohort was 1.13 (0.41e2.46), based on six deaths [165], and no NHL cases were observed in the German workers with chloracne [174]. In summary, a substantial proportion of manufacturers and sprayers included in these cohorts experienced considerably higher exposure to TCDD than the general population, as demonstrated by the serum TCDD levels measured in several subsets of workers. Notwithstanding the complications from the overlap among reports, somedbut not mostdstudies of manufacturers and sprayers of TCDD-contaminated herbicides show a significant positive association with NHL mortality. Limitations of the available data include the fact that most studies reported only results for NHL mortality and not incidence, that no study reported results for specific subtypes of lymphoma, and that the number of workers with high documented exposure was relatively small. In addition, exposure assessment was indirectly estimated based on employment history or extrapolated biomonitoring data, and workers with presumably different levels of TCDD exposure were combined for analysis, resulting in some degree of exposure misclassification. Although exposure misclassification was probably independent from disease status in most studies, it may not have been independent from other classification errors and thus may not have met the conditions necessary for average bias of estimates toward the null [181]. Even under such conditions, any given estimate may be away from the null; thus, underestimated associations cannot be assumed. Results for HL are sparse and do not show an association with TCDD exposure. Results for MM are limited by small numbers of deaths in most studies and lack of exposure-response analyses. An excess of MM mortality was reported in the main analysis of the extended NIOSH cohort [165] and the cohort from New Zealand [171,173] but not in the pooled IARC/NIOSH cohort [164] or subsequent updates of several US cohorts [168,169,172]. Studies of the Seveso industrial accident An accident in a trichlorophenol production plant occurred in 1976 in Seveso, Italy, resulting in exposure to TCDD in part of the population living in six Italian municipalities. This population has been involved in a long-term surveillance study of cancer incidence and mortality. The two most comprehensive reports are based on follow-up for cancer mortality through 2001 [182] and cancer incidence through 1996 [183]. The results of these studies are summarized in Supplementary Table 4. The study included three exposed groups: those living at the time of the accident in the area with the highest exposure (zone A, median serum TCDD level in 1976 ¼ 447.0 ppt; n ¼ 723), those living in an area with intermediate exposure (zone B, median serum TCDD ¼ 94.0 ppt; n ¼ 4821), and those living in an area with low

284

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

exposure (zone R, median TCDD level ¼ 48.0 ppt; n ¼ 31,643). Those living in the remaining areas of the six municipalities and the population in the five surrounding municipalities (n ¼ 181,574) comprised the reference group. Incident cancer cases were ascertained from record linkage with the regional population-based cancer registry, and the cause of death was obtained by linkage to the National Central Statistics Institute and Lombardy regional local health units or from the regional statistical office. The analysis of NHL incidence [183] included one case in zone A group (RR ¼ 0.80 [0.11e5.69]), 12 cases in zone B group (RR ¼ 1.51 [0.85e2.69]), and 49 cases in zone R group (RR ¼ 0.90 [0.66e1.22]; Supplementary Table 4). The only case in zone A occurred 15 years or more after the accident (RR in this category ¼ 1.97 [0.27e14.07]). RRs in zone B were 1.75 (0.43e7.20; 2 cases) 0 to 4 years after the accident, 1.37 (0.34e5.62; 2 cases) 5 to 9 years after, 2.30 (0.93e5.66; 5 cases) 10 to 4 years after, and 0.94 (0.30e2.96; 3 cases) 15 years or more after. A similar analysis in zone R revealed no apparent trend by time since the accident. No cancer cases developed among the 183 Seveso residents who were diagnosed with chloracne shortly after the accident (1.7 cases expected). The analysis of HL incidence included no cases in zone A, three cases in zone B (RR ¼ 1.20 [0.38e3.78]), and 23 cases in zone R (RR ¼ 1.46 [0.91e2.29]). Results by time since the accident were not reported. The one case of MM that occurred in zone A (RR ¼ 2.88 [0.40e20.70]) was diagnosed 15 years or more after the accident (RR in this category ¼ 8.35 [1.14e61.31]). In zone B, six cases of MM were diagnosed (RR ¼ 2.77 [1.2e6.32]). In the analysis by time since the accident within zone B, there were two cases in the 0- to 4-year category (RR ¼ 3.56 [0.85e15.00]), two cases in the 5- to 9-year category (RR ¼ 4.76 [1.11e20.38]), two cases in the 10- to 14-year category (RR ¼ 4.47 [1.04e19.20]), and no cases in the 15 years category. In zone R, 18 cases of MM were observed (RR ¼ 1.15 [0.70e1.91]), 11 of which occurred 15 years or more after the accident (RR in that category ¼ 2.24 [1.11e4.49]). The cancer mortality analysis included also those individuals who were born or immigrated into the study area between 1976 and 1985 [182]. Compared with the reference group, RRs for NHL were 3.35 (1.07e10.46; 3 deaths) in zone A, 1.23 (0.58e2.60; 7 deaths) in zone B, and 0.99 (0.71e1.38; 4 deaths) in zone R. The RR among women in zone A was 4.45 (1.10e17.99; 2 deaths; Supplementary Table 4). There were no HL deaths in zone A; RRs in zones B and R were 2.15 (0.67e6.86; 3 deaths) and 0.94 (0.46e1.89; 9 deaths), respectively. RRs for MM were 4.34 (1.07e17.52; 2 deaths) in zone A, 1.68 (0.69e4.10; 5 deaths) in zone B, and 1.10 (0.71e1.69, 24 deaths) in zone R. The RR of MM among women in zone B was 3.07 (1.12e8.42), based on five deaths. Results according to time since the accident were not reported for specific types of lymphopoietic neoplasms. This study did not use incidence-based mortality; rather, it included deaths from neoplasms diagnosed before the accident that would be uninformative about any association with TCDD exposure. However, bias should have been minimal, if any, because deaths from preaccident neoplasms were included from all zones. A separate study was conducted on 981 women who were residing in zones A and B, were aged 0 to 40 years at the time of the accident, had a serum sample collected close to the time of the accident, and agreed in 1996 to participate in the study (80% participation rate) [184]. In 2008, these women (excluding 16 who had died and 36 who could not be contacted) were recontacted, and 833 (84.9%) who agreed to participate were interviewed. For women who reported in the interview that they were diagnosed with cancer, past medical records were obtained and were reviewed by a pathologist. Death certificates were obtained for deceased women. TCDD was measured in the archived serum

samples by high-resolution gas chromatography and mass spectrometry. Two cases of lymphoma (coded as NHL) were reported in this study. The expected numbers of cases were not provided, nor were analyses according to measured TCDD level reported for these neoplasms. In summary, there is some evidence of increased risk of NHL and MM in the community exposed to the Seveso accident with followup extending for 20 to 25 years. Although the number of individuals in the zone with highest exposure is small, and exposure was classified with some error based on geographic zone of residence instead of individual serum TCDD level, the results in the combined zones A and B show an association of TCDD exposure with NHL mortality and MM incidence and mortality. The increase in NHL mortality but not incidence in zone A suggests an inconsistency that could be due in part to misclassification of the cause of death on death certificates, such that incidence data would provide more valid results [100]. Studies of hospitalized patients and community members with measured TCDD exposure Hospital- and community-based studies that did not focus on groups with potential military or industrial exposure to Agent Orange or TCDD offer limited additional information about the association with risk of NHL. The results of the four studies in this group, none of which estimated associations with risk of NHL subtypes, HL, or MM, are summarized in Supplementary Table 5. In a small cross-sectional study in Sweden, TCDD and other polychlorinated dibenzo-p-dioxins and dibenzofurans were measured in abdominal wall adipose tissue collected from seven histopathologically confirmed NHL cases and one HL case seen at the Umeå University Hospital oncology department in the period from 1989 to 1991, along with 12 elective surgery patients seen at the same hospital [185,186]. Lymphoid malignancy cases were selected to have a history of probable or potential occupational or dietary exposure to dioxins and dibenzofurans, whereas controls were selected not to have such an exposure history. The mean adipose tissue TCDD concentration did not differ significantly between NHL cases (mean ¼ 10.1 pg/g wet basis; range, 0.4e36) and controls (mean ¼ 2.8 pg/g wet basis; range, 0.4e6; P ¼ .22; Supplementary Table 5) [185,186]. The adipose tissue TCDD concentration in the HL case was 2.0 pg/g wet basis. However, mean adipose tissue TCDD TEQ calculated from measured dioxins and dibenzofurans as a measure of aryl hydrocarbon (Ah) receptoremediated signaling disruption was significantly higher among NHL cases (mean ¼ 64.7 pg/g wet basis; range ¼ 19.9e187) than controls (mean ¼ 29.7 pg/g wet basis, range ¼ 12.9e53.4; P ¼ .04) and lower in the HL case (14.0 pg/g wet basis). Given that the study was limited to probably or potentially exposed cases and unexposed controls, the results do not provide useful information about the association between TCDD and NHL or HL risk. Moreover, the findings may have been affected by reverse causality because TCDD levels were measured after NHL diagnosis and possibly treatment, both of which could increase adipose tissue levels of dioxins and dibenzofurans because of weight loss. A subsequent case-control study compared 33 consecutive patients with newly diagnosed, histopathologically confirmed, untreated NHL and 39 consecutive surgery patients with benign lesions from the oncology and surgery departments, respectively, at Örebro Medical Center in Sweden [187,188]. Participation rates were 100% in both groups. No subjects were classified as being occupationally exposed to dioxins or dibenzofurans. Again, TCDD and other polychlorinated dibenzo-p-dioxins and dibenzofurans, as well as polychlorinated biphenyls, were measured in abdominal wall adipose tissue. In addition, antibodies against the Epstein-Barr

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

virus (EBV) viral capsid antigen, restricted and diffuse early antigen, and nuclear antigen 1 were measured in 23 cases and 22 controls. No significant difference in the distribution of adipose tissue TCDD concentration was detected between cases (mean ¼ 3.2 pg/g lipid basis; median ¼ 3.0 pg/g; range ¼ 1.0e5.8 pg/g) and controls (mean ¼ 3.3 pg/g lipid basis, median ¼ 3.2 pg/g, range ¼ 0.92e9.5 pg/g; Wilcoxon P ¼ .85; Supplementary Table 5) [187,188]. After cross-classification by adipose tissue, TCDD concentration (>or 3.2 pg/g lipid basis) and anti-EBV early antigen immunoglobulin G (IgG) titer (>or 80), no significant difference was detected between those with lower levels of both exposures (reference group; 7 NHL cases) and those with elevated TCDD only (OR ¼ 0.45 [0.07e2.4]; 3 NHL cases) or those with elevated levels of both TCDD and anti-EBV early antigen IgG (OR ¼ 0.90 [0.19e4.2; 6 NHL cases). Likewise, no significant association was detected when the analysis was restricted to low-grade or high-grade B-cell NHL. When adipose tissue TEQ (calculated on the basis of TCDD, seven other dioxins, and 10 furans) was used as the exposure metric, the distribution did not differ significantly between cases and controls (Wilcoxon P ¼ .63). NHL risk was higher, but not significantly so, in those with elevated TEQ (>27.79 pg/g lipid basis) and elevated antiEBV early antigen IgG (>80) compared with those who had lower levels of both (Supplementary Table 5). Although analyses were adjusted for body mass index at the time of sampling or one year before [187], this adjustment may not have been sufficient to control for an effect of NHL-related weight loss on adipose tissue TCDD concentrations. However, the expected bias would be toward a positive association [189]. Despite other limitations, including small size, possibly different case and control source populations, and a relatively narrow range of TCDD concentrations, the results of this study suggest no association between long-term cumulative TCDD exposure, measured on the basis of adipose tissue concentration, and NHL risk. A hospital-based case-control study initiated in 1993 in Ho Chi Minh City enrolled 150 NHL cases and approximately 300 controls (half with cancers other than cancer of the liver, respiratory organ, prostate, or thyroid and other endocrine glands, HL, leukemia, or MM, and half with nonmalignant conditions, mostly benign tumors) matched to cases on age, sex, and place of residence, including 124 complete matched triplets [190,191]. All subjects were requested to provide blood and adipose tissue specimens for TCDD measurement, and potential exposure to Agent Orange was evaluated based on self-report and the Exposure Opportunity Index model developed by Stellman and Stellman [192] that incorporated self-reported residential history and U.S. Army herbicide spraying records. Full results from this study have not been published, but a 2004 conference abstract provided preliminary results, pending “refinement of the [Exposure Opportunity Index] calculations” [191]. According to these results, self-reported recall of spraying missions during the war was not significantly associated with NHL risk whether NHL cases were compared with both sets of controls (OR ¼ 1.2 [0.57e2.56]; 13 cases) or either cancer or noncancer controls (Supplementary Table 5) [191]. Having served in the army during the Vietnam War (OR among males ¼ 0.72 [0.36e1.47]; 19 cases) and having an Exposure Opportunity Index at or above the 90th percentile (OR ¼ 1.1 [0.52e2.18]; 13 cases) also were not significantly associated with NHL risk in any of the three comparisons tested. However, having an Exposure Opportunity Index above the median (OR vs. all controls ¼ 1.4 [0.89e2.10]; 57 cases) was significantly associated with NHL risk in comparison with noncancer controls (OR ¼ 2.2 [1.18e4.09]), and a statistically significant trend was detected across tertiles of the Exposure Opportunity Index in the latter comparison (Ptrend ¼ .040). No significant association was detected when comparing NHL cases

285

with cancer controls (Ptrend ¼ .824) or both sets of controls (Ptrend ¼ .298). The authors stated that “a pilot analysis conducted on 25 [adipose tissue samples] did not show a good correlation between the measured tissue dioxin levels and the calculated [Exposure Opportunity Index]” [191]. Therefore, the interpretation of the association between the Exposure Opportunity Index and NHL compared with noncancer controls is unknown, but it is unlikely to reflect an association between TCDD exposure and NHL risk. Combined with concerns about possible selection bias (participation rates were not reported; 17% of the 150 NHL cases were excluded from analyses) and confounding due to the apparent lack of control for covariates other than the matching factors, the absence of valid exposure estimates in this study renders the results largely uninformative about the causal association between Agent Orange or TCDD exposure and NHL risk. Finally, plasma concentrations of TCDD, six other dioxins, 10 furans, 36 noncoplanar and four coplanar polychlorinated biphenyl congeners, and 13 organochlorine pesticides or metabolites were measured in a population-based case-control study of 100 untreated NHL cases and 100 controls from four U.S. regions enrolled between 1998 and 2000 [193]. This study was based on a larger case-control study of NHL cases identified from population-based cancer registries and controls identified by random-digit dialing (if aged < 65 years) or Medicare eligibility files (if aged  65 years), with participation rates of 59% for cases and 44% for controls. Plasma TCDD was detected in two cases and five controls (median limit of detection ¼ 4.1 pg/g lipid); the authors did not estimate an OR for detectable TCDD (Supplementary Table 5). Overall plasma TEQ was positively associated with NHL risk, with an OR of 1.35 (1.02e1.79) per 10 pg/g lipid increase in TEQ. However, TEQ from dioxins was not significantly associated with NHL risk, nor was the summed molar concentration of all measured dioxins. Limitations of this study [193] include the relatively high limits of detection, the possibility of selection bias in an unknown direction (e.g., if occupation or diet, which can affect the exposures of interest, were related to participation and to NHL risk), and the potential influence of NHL symptoms on plasma levels of TCDD and other analytes. Given that TCDD was largely undetectable among study subjects, this study provides scant information about the association between TCDD and NHL risk, other than to suggest that background levels of TCDD probably play little to no role in NHL development in general populations. In summary, studies of communities and hospitalized subjects not selected for military or industrial exposure to Agent Orange or TCDD do not demonstrate any significant association between valid measures of TCDD exposure and risk of NHL. Only one of the four studies [191] found a significant positive association between an elevated Agent Orange Exposure Opportunity Index and NHL risk in some but not all analyses, but the poor correlation of this index with adipose tissue TCDD levels indicates that the index is probably invalid. This interpretation is supported by the lack of significant or otherwise remarkable differences in TCDD levels between NHL cases and controls in other community- or hospital-based studies. Ecologic studies with residential proximity to a TCDD source Ecologic studies, which compare groups rather than individuals, provide limited information for causal inference, in large part because individual-level exposures may differ substantially from average group-level exposures and because associations that are observed for groups may not hold for individuals [194]. Three ecologic studies were identified that estimated TCDD exposure based on residential proximity to a TCDD source [195e197]. These

286

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

studies are described briefly here, and details are provided in Supplementary Table 7. In a study of cancer mortality in the period from 1995 to 1998 among residents of Chapaevsk, Russia, the site of a chemical plant that produced chemical blister agents before 1949, hexachlorocyclohexane (lindane) from 1967 to 1987, and various chlorinated crop protection chemicals thereafter, no significant excess of lymphoma mortality was detected (SMR for males ¼ 1.1 [0.4e2.3], 6 deaths; SMR for females ¼ 1.1 [0.5e2.1], 9 deaths) [197]. Among residents living within 1.5 km of a former chemical manufacturing plant in Rhodes, New South Wales, Australia, that produced creosote, xanthates, dichlorodiphenyltrichloroethane, 2,4-dichlorophenoxyacetic acid, and 2,4,5-T from 1927 to 1986, 54 lymphoid malignancies (63.5% of all hematopoietic malignancies, including 40 cases of NHL, 9 cases of MM, and 5 other cases) were diagnosed in the period from 1972 to 2001 [195]. SIRs and SMRs were not reported for specific lymphoid malignancies, but the SIRs and SMRs for hematopoietic neoplasms were all below 1.0, the SIRs significantly so, in comparison with the state and the local region. A study of residents living around a former agrochemical plant that manufactured 2,4,5-T from 1962 to 1987 found a significantly higher than expected incidence of NHL and CLL in the period from 1970 to 1974 (NHL SIR ¼ 1.75 [1.21e2.46] based on 33 cases; CLL SIR ¼ 2.51 [1.44e4.08] based on 16 cases), a significant increase in NHL mortality in the period from 1975 to 1979 (SMR ¼ 1.58 [1.01e2.35]; 24 deaths), and a significant increase in CLL incidence in the period from 1980 to 1984 (SIR ¼ 2.55 [1.58e3.91]; 21 cases) [196]. Otherwise, NHL, CLL, and HL incidence and mortality rates did not differ significantly from expectation between 1970 and 2001. The authors noted that the period from 1970 to 1974, when both NHL and CLL incidence were higher than expected, was partly outside a 10-year latency period after the plant started manufacturing 2,4,5-T, and that annual production was lower from 1962 to 1964 than from 1969 to 1972. Altogether, these studies do not demonstrate a consistent excess of NHL, HL, or MM in communities residing near an industrial source of TCDD. Given the lack of individual-level information on TCDD exposure, the temporal ambiguity between exposure and outcome, the inability to account for duration of residence, and the poor control for confounding, these ecologic studies provide minimal evidence for or against a relationship between actual exposure to TCDD and risk of lymphoid malignancies. Weight-of-Evidence evaluation Strength Most estimates for the association between Agent Orange or TCDD exposure and risk of NHL, HL, or MM incidence or mortality were between 0.3 and 3.0. The stronger RRs detected in some studies were mostly based on fewer than five cases and therefore statistically unstable [120,167,169,171e173,180,182,183] or, in the case of one study, based on an invalid and almost certainly biased self-reported exposure estimate [126]. In addition, bias and confounding were not completely addressed in all studies. Across all studies, the only RR above 3.0 that was based on five or more cases was the SMR of 4.85 (2.33e8.93; 10 deaths) for NHL among workers with “other exposed jobs” in the pooled IARC/NIOSH cohort [164]. No other job groups in that cohort, including main production, maintenance, unspecified tasks, and spraying, had a significant excess of NHL mortality. Overall, the strength of observed associations does not provide compelling evidence in favor of a causal relationship between Agent Orange or TCDD and NHL, HL, or MM, although a causal effect cannot be excluded on the basis of weak associations.

Consistency Selected significant positive associations with NHL incidence or mortality were detected in some studies [167,169,172,182] (only for self-perceived exposure in [126], only in the comparison with cancer controls in [191], and only for <1 year of exposure or “other exposed jobs” in [164]) but not in most studies [114,120,123,127,165,166,168,171,173,174,180,183,186e188]. Most of the statistically nonsignificant associations in the industrial cohorts were in a positive direction. However, because these studies could not reasonably rule out confounding or bias, especially in light of the observed strength of associations, it is unpredictable whether the estimated RRs would have been the same but statistically significant if the studies had been larger in size. Therefore, the nonsignificant RRs above 1.0 are interpretable as being consistent with either a positive association or no association. Across all study settings, associations with NHL risk were not consistent. For HL, a significant positive association was observed only with self-perceived Agent Orange exposure in one study [126]; otherwise, all associations were statistically null [114,120,122,127, 164,165,168,169,171e173,182,183]. Estimated exposure to Agent Orange or TCDD was significantly (or nearly significantly) associated with increased MM incidence or mortality in some studies [126,165,173,182,183], including both of the Seveso cancer studies, but not in others [114,127,164,171,172]. Taken together, associations between exposure to Agent Orange or TCDD and risk of NHL, HL, or MM were not consistently detected among or within studies. However, for NHL the industrial cohorts, in which TCDD exposures were the highest, offer more consistent evidence of a positive association, albeit not always statistically significant. For MM, the number of available studies is limited, but among these a positive association was found in a sizable proportion of high-quality, relevant studies. Specificity IARC classifies TCDD as carcinogenic for all cancers combined in humans, and identifies “a positive association” with NHL (and softtissue sarcoma and lung cancer) specifically [20]. TCDD is the only agent classified by IARC as an all-site carcinogen [198], but the concept that a single chemical could cause all cancers is of questionable validity [199]. IARC’s original classification of TCDD as an all-site carcinogen was based in large part on observations that TCDD causes several cancer types in experimental animals through a tumor-promotion mechanism involving the Ah receptor and that this receptor and related signaling pathways and responses are highly conserved across species, including humans [20]. Temporality Most studies discussed in this review were designed to ascertain NHL, HL, and MM incidence or mortality after potential exposure to Agent Orange or TCDD, whether through military service in Vietnam, employment in chlorophenoxy herbicide manufacture or spraying, or proximity to the Seveso accident. The Air Force Health Study measured serum TCDD levels in 1987 or later [107], possibly after the onset of lymphoid malignancy in some subjects. However, the persistently higher serum TCDD levels in Ranch Hand veterans than comparison subjects [108,109] suggest that participation in Agent Orange aerial spraying missions during the Vietnam War, before the onset of cancer, was the primary exposure source. The fact that exposure preceded the outcomes of interest in these studies make their results temporally consistent with a causal association (or lack thereof). By contrast, hospital- and communitybased studies were unable to establish the temporal sequence

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

between biologically measured TCDD levels and the onset of disease [185e188,190,191,193], rendering their results less useful for causal inference. Of note, the latency period between Agent Orange or TCDD exposure and development of lymphoid malignancies is essentially unknown and possibly variable given that TCDD is presumed to exert a carcinogenic effect through a nongenotoxic mechanism. It is worth noting that for some mechanisms implicated in lymphomagenesis, notably immunosuppression after organ transplantation, the induction time is short, with the highest excess risks occurring within one year [200]. Biological gradient Only the industrial cohort studies and the Seveso study had sufficient exposure data and subject numbers to evaluate exposureresponse trends in risk of lymphoid malignancies across increasing categories of exposure to TCDD, and most studies evaluated trends only for NHL. In several studies, no apparent or statistically significant trend was detected between increasing exposure and risk of NHL mortality. In the international IARC/NIOSH pooled analysis, NHL mortality did not increase with longer duration of exposure, and the SMR for NHL was not significantly higher among production workers than sprayers despite the substantially higher serum TCDD levels measured in the former group [164]. The nested casecontrol study of NHL within the IARC cohort found higher RRs in the medium and high categories of estimated TCDD exposure than the nonexposed and low-exposure categories, but no statistically significant trend was detected [180]. Among U.S. trichlorophenol manufacturers, no significant linear association was detected between estimated cumulative TCDD exposure and NHL mortality [168]. In the study of New Zealand trichlorophenol manufacturers, NHL mortality did not rise in concert with increasing categories of estimated cumulative TCDD exposure [171]. Among workers at four U.S. pentachlorophenol production plants, longer duration of work was not associated with greater risk of NHL mortality [172]. However, in a separate analysis at one of these plants, the SMR for NHL increased across three categories of estimated cumulative TCDD exposure (albeit perhaps not significantly; no test for trend was reported, and the SMR was statistically nonsignificant in all categories), whereas no significant trends were detected for increasing categories of estimated cumulative TEQ or continuous cumulative TEQ [169]. A significant log-linear association and monotonically increasing risk across tertiles was detected between plasma TCDD level and NHL mortality at two Dutch chlorophenoxy herbicide plants [167]. Furthermore, in the Seveso community, risk of death from both NHL and MM increased progressively across zones R, B, and A compared with the reference zone [182]. No such trend was detected for HL mortality, as no HL deaths occurred among zone A residents, although risk was higher in zone B than zone R. However, in the analysis of cancer incidence in Seveso, no apparent exposure-response trend was detected between residential zone and risk of either NHL or HL [183]. For MM incidence, risk increased progressively across zones R, B, and, A, with a significant excess only in zone B. Other studies that evaluated exposure-response trends were uninformative about the biological gradient between actual Agent Orange or TCDD exposure and risk of lymphoid malignancies. In the studies of cancer incidence and prevalence in Korean veterans, no significant trends with risk of NHL, HL, or MM were consistently detected across categories of estimated Agent Orange exposure [126,127]. However, exposure estimates were not significantly correlated with measured serum TCDD levels among Korean veterans [129]. Likewise, in the Vietnam hospital-based case-control study, a significant positive trend was detected for NHL risk, in

287

comparison with noncancer controls, across tertiles of the Agent Orange Exposure Opportunity Index, but this estimate was shown not to be a valid estimate of TCDD exposure [191]. No significant trend was detected between quartiles of plasma dioxin TEQ or total dioxin molar concentration and NHL risk in the U.S. populationebased case-control study [193], but this study did not have a sufficient number of subjects with detectable plasma TCDD to analyze this exposure specifically. Taken together, the results from occupational cohorts and Seveso do not demonstrate a consistent monotonic trend between increasing TCDD exposure and NHL mortality, even at relatively high exposure levels. Evidence regarding trends for other outcomes of interest is derived only from the Seveso study, in which no clear trend was observed for NHL incidence, HL incidence, or HL mortality, but positive trends were apparent for MM incidence and mortality. However, independent confirmation of these findings is needed from other studies. Therefore, available evidence for a biological gradient between exposure to Agent Orange or TCDD and risk of lymphoid malignancies is inconclusive and does not plainly support or detract from a causal conclusion. Plausibility TCDD binds to and activates the Ah receptor, thereby causing changes in gene expression, cell replication, apoptosis, and oxidative stress that may promote or indirectly initiate carcinogenesis [20]. In vitro studies suggest that the Ah receptor plays an important role in B- and T-cell development and function [201,202]. TCDD could potentially also increase the risk of lymphoid malignancies through an Ah receptoremediated mechanism involving immune suppression [203,204]. However, whether occupational exposure to Agent Orange or TCDD can cause the extreme levels of immune suppression known to increase NHL risk (i.e., levels comparable with those found in HIV infection or organ transplantation) is unknown. In light of the limited understanding of differences in the pathogenic mechanisms that lead to NHL, HL, MM, and various subtypes of these diseases, no direct evidence is available to support these hypotheses in the development of specific lymphoid neoplasms in humans. Coherence Histiocytic lymphomas were increased in female B6C3F1 mice (but not male B6C3F1 mice, other mouse strains, or rats) exposed to TCDD by oral gavage, and thymic and nonthymic lymphomas were increased in immature male and female B6C3 and B6C mice (but not hamsters) exposed to TCDD by intraperitoneal injection [20,205]. Topical skin application of TCDD in mice and subcutaneous injection in hamsters did not produce lymphomas. The inconsistency of results for lymphomagenesis across species, strains, and routes of exposure diminishes the coherence of the available scientific evidence for a causal effect of TCDD on NHL, HL, and MM in humans. Furthermore, the tumors that are most frequently increased in experimental animals exposed to TCDD are squamous cell carcinomas of the skin and mucosa and hepatocellular adenomas and carcinomas, which do not correspond to lymphoid malignancies. However, the lack of site concordance between animals and humans does not detract from the evidence for a causal effect of TCDD on NHL, HL, and MM in humans; instead, it simply does not augment such evidence. Experiment Experimental evidence is lacking with regard to whether TCDD exposure causes lymphoid malignancies in humans or whether

288

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

removing TCDD from an exposed population prevents the development of lymphoid malignancies. The incidence of NHL increased significantly in the United States in the late 20th century and the incidence of MM continues to rise [206], despite the Environmental Protection Agency’s series of bans on uses of 2,4,5-T and other chlorophenoxy herbicides in the 1970s and 1980s [207]. Moreover, NHL, HL, and MM incidence rates are not consistently higher (and, in most instances, are substantially lower) in Southern Vietnam (Ho Chi Minh City) than in Northern Vietnam (Hanoi) and other regions in Southeast Asia where Agent Orange was not used [208,209]. However, these ecologic data do not constitute strong evidence for or against a causal effect of Agent Orange or TCDD on lymphoid malignancies. Analogy No chemical agents are firmly established as causes of NHL, HL, or MM. Conclusions In summary, the weight of epidemiologic evidence shows that TCDD exposure is not consistently associated with increased risk of NHL mortality. A significant positive association was observed in several studies of herbicide manufacturers and sprayers and one study of NHL mortality (but not incidence) in the Seveso community. Although such an association is temporally logical and biologically plausible, the strength of the observed association and the lack of a consistent positive biological gradient do not provide conclusive evidence to support a causal interpretation. Among Vietnam veterans and community members, the weight of epidemiologic evidence does not demonstrate a positive association between Agent Orange or TCDD exposure and increased NHL risk. In addition, most studies showed no significant difference in NHL risk between Vietnam veterans and non-Vietnam veterans or the general population. Average serum TCDD levels from military exposure to Agent Orange and those observed in general community members are substantially lower than those measured among exposed sprayers, especially manufacturers of TCDD-contaminated herbicides, and among Seveso residents. This pattern of findings could suggest an exposure-response trend with increased NHL risk at very high levels of exposure to TCDD but such a trend requires more direct investigation to be confirmed. Of note, mortality from lymphatic and hematopoietic malignancies combined was not significantly elevated among workers in the NIOSH cohort who developed chloracne, which suggests high exposure to TCDD [165]. For MM, the combined epidemiologic evidence on the association with Agent Orange or TCDD exposure is suggestive but inconclusive. The significant positive association detected in some studies of occupational or accidental community exposure to TCDDcontaminated herbicides is temporally logical, biologically plausible, and in line with a positive exposure-response gradient. However, such results are based on a small number of studies: five with a significant positive association, including twodboth in Sevesodthat demonstrated a positive exposure-response trend with an informative estimate of TCDD exposure. The few available studies of Vietnam veterans did not provide strong evidence for or against an excess of MM in association with estimated Agent Orange exposure or Vietnam military service in general. More robust results from additional study settings are needed to confirm whether TCDD exposure is associated with increased MM risk. Overall, the epidemiologic evidence does not demonstrate a strong or consistent association or an exposure-response trend between Agent Orange or TCDD exposure and HL risk. Studies of

TCDD-contaminated herbicide manufacturers and sprayers, Seveso residents, and Vietnam veterans are in accordance in not demonstrating a clear excess of HL risk. Our conclusions do not necessarily disagree with those of the IOM Committee, whichdas described in the introduction sectiondis charged with determining whether a statistical association (not a causal association) exists between specific health outcomes and exposure to TCDD and other chemicals in herbicides used by the military in Vietnam [10]. The IOM Committee has determined that a positive statistical association exists between herbicides and NHL and HL, whereas the evidence of such an association with MM is “limited or suggestive” [10e19]. Similar to the IOM Committee, we identified several studies in which a statistically significant positive association was detected between estimated TCDD exposure and risk of NHL and HL, and also MM. In particular, such associations were reported in some studies of manufacturers and sprayers of TCDD-contaminated herbicides and in some studies of the Seveso community. However, the scope of our review extends beyond that of the IOM Committee in that we conclude, based on the collective evidence, that a causal association has not been established between TCDD and any of the lymphoid malignancies. A major challenge in epidemiology is disentangling associations with different subtypes of NHL and HL, and perhaps also MM. Indeed, evidence of etiologic heterogeneity among NHL subtypes is growing [210]. Among all the studies discussed in this review, none estimated associations of Agent Orange or TCDD with specific histopathologic subtypes of NHL (other than CLL, which was analyzed as a separate disease entity in the Air Force Health Study) or HL. Four studies estimated associations of Vietnam military service (not specific to Agent Orange exposure) with CLL [139,145,146,149], but only one of these also analyzed other NHL subtypes [146]. Thus, whether associations with Agent Orange or TCDD exposure vary among lymphoid malignancy subtypes, and in particular, whether a positive association with a specific subtype is obscured by the analysis of combined disease entities, is yet unknown. The restriction of the industrial cohort studies to the evaluation of lymphoid cancer mortality, rather than incidence, also limits inference about whether TCDD is associated with disease risk or only prognosis, or whether associations are restricted to more lethal subtypes. The IOM has concluded that because of the shortage of valid quantitative exposure data, “it is very unlikely that additional information or more sophisticated methods are going to become available that would permit any sort of quantitative assessment of Vietnam veterans’ increased risks of particular adverse health outcomes attributable to exposure to the chemicals associated with herbicide spraying in Vietnam” [10]. Instead, continued follow-up of the industrial and Seveso cohorts and larger community-based studies with serum TCDD data may be the best available sources of future evidence to resolve the question of whether TCDD causes NHL, HL, MM, or any specific subtypes of these lymphoid malignancies. Acknowledgments Dr. Chang is employed by Exponent, Inc., a for-profit corporation that provides engineering and scientific consulting services. Dr. Mandel was employed by Exponent, Inc., when this article was initially submitted for publication but has since retired. All the authors have consulted with private or government organizations on the health effects of environmental and occupational exposures, including Agent Orange and TCDD, and on malignant lymphomas. This independent scientific review was financially supported by the Dow Chemical Company and Monsanto Company. The content of this article is the sole responsibility of the authors and does not

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

necessarily represent the views or opinions of Dow Chemical Company or Monsanto Company or any of the employers of the authors. As the study sponsors, Dow Chemical Company and Monsanto Company did not influence the methods or materials used by the authors in conducting their independent review and synthesis of the epidemiological literature or the interpretation of the results, nor were they involved in the preparation or review of the article. The study sponsors did not review or provide feedback on this article before its submission.

References [1] Swerdlow SH, Campo EH, Harris NL, Jaffe ES, Pileri SA, Stein H, et al., editors. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. 4th ed. Geneva: World Health Organization (WHO); 2008. [2] Dreiher J, Kordysh E. Non-Hodgkin lymphoma and pesticide exposure: 25 years of research. Acta Haematol 2006;116(3):153e64. [3] Cocco P, Satta G, Dubois S, Pili C, Pilleri M, Zucca M, et al. Lymphoma risk and occupational exposure to pesticides: results of the Epilymph study. Occup Environ Med 2013;70(2):91e8. [4] Chiu BC, Blair A. Pesticides, chromosomal aberrations, and non-Hodgkin’s lymphoma. J Agromedicine 2009;14(2):250e5. [5] Khuder SA, Mutgi AB, Schaub EA, Tano BD. Meta-analysis of Hodgkin’s disease among farmers. Scand J Work Environ Health 1999;25(5):436e41. [6] McCunney RJ. Hodgkin’s disease, work, and the environment. A review. J Occup Environ Med 1999;41(1):36e46. [7] Khuder SA, Mutgi AB. Meta-analyses of multiple myeloma and farming. Am J Ind Med 1997;32(5):510e6. [8] Merhi M, Raynal H, Cahuzac E, Vinson F, Cravedi JP, Gamet-Payrastre L. Occupational exposure to pesticides and risk of hematopoietic cancers: meta-analysis of case-control studies. Cancer Causes Control 2007;18(10): 1209e26. [9] Perrotta C, Staines A, Cocco P. Multiple myeloma and farming. A systematic review of 30 years of research. Where next? J Occup Med Toxicol 2008;3:27. [10] Institute of Medicine. Veterans and Agent Orange: Update 2012. Committee to Review the Health Effects in Vietnam Veterans of Exposure to Herbicides (Ninth Biennial Update). Washington, DC: National Academies Press; 2014. [11] Institute of Medicine. Veterans and Agent Orange: Health Effects of Herbicides Used in Vietnam. Committee to Review the Health Effects in Vietnam Veterans of Exposure to Herbicides. Washington, DC: National Academies Press; 1994. [12] Institute of Medicine. Veterans and Agent Orange: Update 1996. Committee to Review the Health Effects in Vietnam Veterans of Exposure to Herbicides. Washington, DC: National Academies Press; 1996. [13] Institute of Medicine. Veterans and Agent Orange: Update 1998. Committee to Review the Health Effects in Vietnam Veterans of Exposure to Herbicides (Second Biennial Update). Washington, DC: National Academies Press; 1999. [14] Institute of Medicine. Veterans and Agent Orange: Update 2000. Committee to Review the Health Effects in Vietnam Veterans of Exposure to Herbicides (Third Biennial Update). Washington, DC: National Academies Press; 2001. [15] Institute of Medicine. Veterans and Agent Orange: Update 2002. Committee to Review the Health Effects in Vietnam Veterans of Exposure to Herbicides (Fourth Biennial Update). Washington, DC: National Academies Press; 2003. [16] Institute of Medicine. Veterans and Agent Orange: Update 2004. Committee to Review the Health Effects in Vietnam Veterans of Exposure to Herbicides (Fifth Biennial Update). Washington, DC: National Academies Press; 2005. [17] Institute of Medicine. Veterans and Agent Orange: Update 2006. Committee to Review the Health Effects in Vietnam Veterans of Exposure to Herbicides (Fifth Biennial Update). Washington, DC: National Academies Press; 2007. [18] Institute of Medicine. Veterans and Agent Orange: Update 2008. Committee to Review the Health Effects in Vietnam Veterans of Exposure to Herbicides (Fifth Biennial Update). Washington, DC: National Academies Press; 2009. [19] Institute of Medicine. Veterans and Agent Orange: Update 2010. Committee to Review the Health Effects in Vietnam Veterans of Exposure to Herbicides (Eighth Biennial Update). Washington, DC: National Academies Press; 2011. [20] IARC. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Volume 100F. A Review of Human Carcinogens: Chemical Agents and Related Occupations. Lyon: International Association for Research on Cancer (IARC); 2012. p. 599. [21] Hanks H. Agent Orange, PTSD and why an activist fights. Available at: http:// timesofsandiego.com/opinion/2014/09/13/agent-orange-ptsd-activist-fights/. Last accessed 19 September 2014. Times of San Diego. 13 September 2014. [22] Relyea K. Bellingham town hall focuses on impacts of Agent Orange on Vietnam War veterans, families. Available at: http://www.bellinghamher ald.com/2014/09/18/3862748/bellingham-town-hall-focuses-on.html. Last accessed 19 September 2014. The Bellingham Herald. 18 September 2014. [23] Haberman C. Agent Orange’s long legacy, for Vietnam and veterans. Available at: http://www.nytimes.com/2014/05/12/us/agent-oranges-long-legacy-forvietnam-and-veterans.html. Last accessed 19 September 2014. The New York Times. 11 May 2014.

289

[24] Hicks J. Study supports claims of Agent Orange exposure after Vietnam War. Available at: http://www.washingtonpost.com/blogs/federal-eye/wp/2014/ 02/26/study-supports-claims-of-agent-orange-exposure-after-vietnam-war/ . Last accessed 19 September 2014. The Washington Post. 26 February 2014. [25] Ferlay J, Soerjomataram I, Ervik M, Dikshit R, Eser S, Mathers C, et al. GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 11 [Internet]. Lyon: International Agency for Research on Cancer (IARC); 2013. Available at: http://globocan.iarc.fr. Last accessed 19 September 2014. [26] Orem J, Mbidde EK, Lambert B, de Sanjose S, Weiderpass E. Burkitt’s lymphoma in Africa, a review of the epidemiology and etiology. Afr Health Sci 2007;7(3):166e75. [27] Iwanaga M, Watanabe T, Yamaguchi K. Adult T-cell leukemia: a review of epidemiological evidence. Front Microbiol 2012;3:322. [28] Ferlay J, Parkin DM, Curado MP, Bray F, Edwards B, Shin HR, et al. Cancer Incidence in Five Continents, Volumes I to IX: IARC CancerBase No. 9. Lyon, France: International Agency for Research on Cancer (IARC); 2010. Available at: http://ci5.iarc.fr/CI5plus/ci5plus.htm. Last accessed 19 September 2014. [29] Bosetti C, Levi F, Ferlay J, Lucchini F, Negri E, La Vecchia C. Incidence and mortality from non-Hodgkin lymphoma in Europe: the end of an epidemic? Int J Cancer 2008;123(8):1917e23. [30] Turesson I, Velez R, Kristinsson SY, Landgren O. Patterns of multiple myeloma during the past 5 decades: stable incidence rates for all age groups in the population but rapidly changing age distribution in the clinic. Mayo Clinic Proc 2010;85(3):225e30. [31] Melbye M, Smedby KE, Trichopoulos D. Chapter 27. Non-Hodgkin lymphoma. In: Adami HO, Hunter D, Trichopoulos D, editors. Textbook of Cancer Epidemiology. 2nd ed. New York: Oxford University Press; 2008. p. 669e93. [32] Alexander DD, Mink PJ, Adami HO, Chang ET, Cole P, Mandel JS, et al. The non-Hodgkin lymphomas: a review of the epidemiologic literature. Int J Cancer 2007;120(Suppl 12):1e39. [33] Hartge P, Wang SS, Bracci PM, Devesa SS, Holly EA. Chapter 46. Non-Hodgkin lymphoma. In: Schottenfeld D, Fraumeni Jr JF, editors. Cancer Epidemiology and Prevention. 3rd ed. New York: Oxford University Press; 2006. p. 898e918. [34] Ekstrom-Smedby K. Epidemiology and etiology of non-Hodgkin lymphomaea review. Acta Oncol (Stockholm, Sweden) 2006;45(3):258e71. [35] Shiels MS, Engels EA. Increased risk of histologically defined cancer subtypes in human immunodeficiency virus-infected individuals: clues for possible immunosuppression-related or infectious etiology. Cancer 2012;118(19): 4869e76. [36] Navarro WH, Kaplan LD. AIDS-related lymphoproliferative disease. Blood 2006;107(1):13e20. [37] Hjalgrim H, Engels EA. Infectious aetiology of Hodgkin and non-Hodgkin lymphomas: a review of the epidemiological evidence. J Intern Med 2008; 264(6):537e48. [38] Engels EA. Infectious agents as causes of non-Hodgkin lymphoma. Cancer Epidemiol biomarkers Prev 2007;16(3):401e4. [39] de Sanjose S, Benavente Y, Vajdic CM, Engels EA, Morton LM, Bracci PM, et al. Hepatitis C and non-Hodgkin lymphoma among 4784 cases and 6269 controls from the International Lymphoma Epidemiology Consortium. Clin Gastroenterol Hepatol 2008;6(4):451e8. [40] IARC. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Volume 100B. A Review of Human Carcinogens: Biological Agents. Geneva: International Agency for Research on Cancer (IARC); 2012. [41] Chao C, Page JH. Type 2 diabetes mellitus and risk of non-Hodgkin lymphoma: a systematic review and meta-analysis. Am J Epidemiol 2008;168(5): 471e80. [42] Mitri J, Castillo J, Pittas AG. Diabetes and risk of Non-Hodgkin’s lymphoma: a meta-analysis of observational studies. Diabetes Care 2008;31(12):2391e7. [43] Cocco P, Vermeulen R, Flore V, Nonne T, Campagna M, Purdue M, et al. Occupational exposure to trichloroethylene and risk of non-Hodgkin lymphoma and its major subtypes: a pooled InterLymph [correction of IinterLlymph] analysis. Occup Environ Med 2013;70(11):795e802. [44] Karami S, Bassig B, Stewart PA, Lee KM, Rothman N, Moore LE, et al. Occupational trichloroethylene exposure and risk of lymphatic and haematopoietic cancers: a meta-analysis. Occup Environ Med 2013;70(8):591e9. [45] Scott CS, Jinot J. Trichloroethylene and cancer: systematic and quantitative review of epidemiologic evidence for identifying hazards. Int J Environ Res Public Health 2011;8(11):4238e72. [46] Rothman N, Skibola CF, Wang SS, Morgan G, Lan Q, Smith MT, et al. Genetic variation in TNF and IL10 and risk of non-Hodgkin lymphoma: a report from the InterLymph Consortium. Lancet Oncol 2006;7(1):27e38. [47] Skibola CF, Bracci PM, Nieters A, Brooks-Wilson A, de Sanjose S, Hughes AM, et al. Tumor necrosis factor (TNF) and lymphotoxin-alpha (LTA) polymorphisms and risk of non-Hodgkin lymphoma in the InterLymph Consortium. Am J Epidemiol 2010;171(3):267e76. [48] Negri E. Sun exposure, vitamin D, and risk of Hodgkin and non-Hodgkin lymphoma. Nutr Cancer 2010;62(7):878e82. [49] Kricker A, Armstrong BK, Hughes AM, Goumas C, Smedby KE, Zheng T, et al. Personal sun exposure and risk of non Hodgkin lymphoma: a pooled analysis from the Interlymph Consortium. Int J Cancer 2008;122(1):144e54. [50] Larsson SC, Wolk A. Body mass index and risk of non-Hodgkin’s and Hodgkin’s lymphoma: a meta-analysis of prospective studies. Eur J Cancer (Oxford, Engl: 1990) 2011;47(16):2422e30.

290

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

[51] Morton LM, Hartge P, Holford TR, Holly EA, Chiu BC, Vineis P, et al. Cigarette smoking and risk of non-Hodgkin lymphoma: a pooled analysis from the International Lymphoma Epidemiology Consortium (interlymph). Cancer Epidemiol biomarkers Prev 2005;14(4):925e33. [52] Morton LM, Zheng T, Holford TR, Holly EA, Chiu BC, Costantini AS, et al. Alcohol consumption and risk of non-Hodgkin lymphoma: a pooled analysis. Lancet Oncol 2005;6(7):469e76. [53] Conde L, Halperin E, Akers NK, Brown KM, Smedby KE, Rothman N, et al. Genome-wide association study of follicular lymphoma identifies a risk locus at 6p21.32. Nat Genet 2010;42(8):661e4. [54] Crowther-Swanepoel D, Broderick P, Di Bernardo MC, Dobbins SE, Torres M, Mansouri M, et al. Common variants at 2q37.3, 8q24.21, 15q21.3 and 16q24.1 influence chronic lymphocytic leukemia risk. Nat Genet 2010;42(2):132e6. [55] Di Bernardo MC, Crowther-Swanepoel D, Broderick P, Webb E, Sellick G, Wild R, et al. A genome-wide association study identifies six susceptibility loci for chronic lymphocytic leukemia. Nat Genet 2008;40(10):1204e10. [56] Kumar V, Matsuo K, Takahashi A, Hosono N, Tsunoda T, Kamatani N, et al. Common variants on 14q32 and 13q12 are associated with DLBCL susceptibility. J Hum Genet 2011;56(6):436e9. [57] Skibola CF, Bracci PM, Halperin E, Conde L, Craig DW, Agana L, et al. Genetic variants at 6p21.33 are associated with susceptibility to follicular lymphoma. Nat Genet 2009;41(8):873e5. [58] Smedby KE, Foo JN, Skibola CF, Darabi H, Conde L, Hjalgrim H, et al. GWAS of follicular lymphoma reveals allelic heterogeneity at 6p21.32 and suggests shared genetic susceptibility with diffuse large B-cell lymphoma. PLoS Genet 2011;7(4):e1001378. [59] Tan DE, Foo JN, Bei JX, Chang J, Peng R, Zheng X, et al. Genome-wide association study of B cell non-Hodgkin lymphoma identifies 3q27 as a susceptibility locus in the Chinese population. Nat Genet 2013;45(7):804e7. [60] Becker N, Falster MO, Vajdic CM, de Sanjose S, Martinez-Maza O, Bracci PM, et al. Self-reported history of infections and the risk of non-Hodgkin lymphoma: an InterLymph pooled analysis. Int J Cancer 2012;131(10):2342e8. [61] Schinasi L, Leon ME. Non-Hodgkin lymphoma and occupational exposure to agricultural pesticide chemical groups and active ingredients: a systematic review and meta-analysis. Int J Environ Res Public Health 2014;11(4): 4449e527. [62] Cross AJ, Lim U. The role of dietary factors in the epidemiology of nonHodgkin’s lymphoma. Leuk Lymphoma 2006;47(12):2477e87. [63] Kane EV, Bernstein L, Bracci PM, Cerhan JR, Costas L, Dal Maso L, et al. Postmenopausal hormone therapy and non-Hodgkin lymphoma: a pooled analysis of InterLymph case-control studies. Ann Oncol 2013;24(2):433e41. [64] Kane EV, Roman E, Becker N, Bernstein L, Boffetta P, Bracci PM, et al. Menstrual and reproductive factors, and hormonal contraception use: associations with non-Hodgkin lymphoma in a pooled analysis of InterLymph case-control studies. Ann Oncol 2012;23(9):2362e74. [65] MacMahon B. Epidemiological evidence of the nature of Hodgkin’s disease. Cancer 1957;10(5):1045e54. [66] MacMahon B. Epidemiology of Hodgkin’s disease. Cancer Res 1966;26(6): 1189e201. [67] Glaser SL, Chang ET, Clarke CA, Keegan TH. Chapter 1. Epidemiology. In: Engert A, Horning SJ, editors. Hematologic Malignancies: Hodgkin Lymphoma A Comprehensive Update on Diagnostics and Clinics. New York: Springer; 2011. p. 3e20. [68] Caporaso NE, Goldin LR, Anderson WF, Landgren O. Current insight on trends, causes, and mechanisms of Hodgkin’s lymphoma. Cancer J (Sudbury, Mass) 2009;15(2):117e23. [69] Cartwright RA, Watkins G. Epidemiology of Hodgkin’s disease: a review. Hematol Oncol 2004;22(1):11e26. [70] Melbye M, Hjalgrim H, Adami HO. Chapter 26. Hodgkin lymphoma. In: Adami HO, Hunter D, Trichopoulos D, editors. Textbook of Cancer Epidemiology. 2nd ed. New York: Oxford University Press; 2008. p. 653e68. [71] Mueller NE, Grufferman S. Chapter 45. Hodgkin lymphoma. In: Schottenfeld D, Fraumeni Jr JF, editors. Cancer Epidemiology and Prevention. 3rd ed. New York: Oxford University Press; 2006. p. 872e97. [72] Glaser SL, Lin RJ, Stewart SL, Ambinder RF, Jarrett RF, Brousset P, et al. Epstein-Barr virus-associated Hodgkin’s disease: epidemiologic characteristics in international data. Int J Cancer 1997;70(4):375e82. [73] Hjalgrim H, Smedby KE, Rostgaard K, Molin D, Hamilton-Dutoit S, Chang ET, et al. Infectious mononucleosis, childhood social environment, and risk of Hodgkin lymphoma. Cancer Res 2007;67(5):2382e8. [74] Hjalgrim H, Askling J, Rostgaard K, Hamilton-Dutoit S, Frisch M, Zhang JS, et al. Characteristics of Hodgkin’s lymphoma after infectious mononucleosis. New Engl J Med 2003;349(14):1324e32. [75] Diepstra A, Niens M, te Meerman GJ, Poppema S, van den Berg A. Genetic susceptibility to Hodgkin’s lymphoma associated with the human leukocyte antigen region. Eur J Haematol Suppl 2005;(66):34e41. [76] Hors J, Dausset J. HLA and susceptibility to Hodgkin’s disease. Immunol Rev 1983;70:167e92. [77] Cozen W, Li D, Best T, Van Den Berg DJ, Gourraud PA, Cortessis VK, et al. A genome-wide meta-analysis of nodular sclerosing Hodgkin lymphoma identifies risk loci at 6p21.32. Blood 2012;119(2):469e75. [78] Enciso-Mora V, Broderick P, Ma Y, Jarrett RF, Hjalgrim H, Hemminki K, et al. A genome-wide association study of Hodgkin’s lymphoma identifies new susceptibility loci at 2p16.1 (REL), 8q24.21 and 10p14 (GATA3). Nat Genet 2010;42(12):1126e30.

[79] Frampton M, da Silva Filho MI, Broderick P, Thomsen H, Forsti A, Vijayakrishnan J, et al. Variation at 3p24.1 and 6q23.3 influences the risk of Hodgkin’s lymphoma. Nat Commun 2013;4:2549. [80] Urayama KY, Jarrett RF, Hjalgrim H, Diepstra A, Kamatani Y, Chabrier A, et al. Genome-wide association study of classical Hodgkin lymphoma and EpsteinBarr virus status-defined subgroups. J Natl Cancer Inst 2012;104(3):240e53. [81] Kamper-Jorgensen M, Rostgaard K, Glaser SL, Zahm SH, Cozen W, Smedby KE, et al. Cigarette smoking and risk of Hodgkin lymphoma and its subtypes: a pooled analysis from the International Lymphoma Epidemiology Consortium (InterLymph). Ann Oncol 2013;24(9):2245e55. [82] Tramacere I, Pelucchi C, Bonifazi M, Bagnardi V, Rota M, Bellocco R, et al. A meta-analysis on alcohol drinking and the risk of Hodgkin lymphoma. Eur J Cancer Prev 2012;21(3):268e73. [83] Monnereau A, Glaser SL, Schupp CW, Ekstrom Smedby K, de Sanjose S, Kane E, et al. Exposure to UV radiation and risk of Hodgkin lymphoma: a pooled analysis. Blood 2013;122(20):3492e9. [84] Martino A, Sainz J, Buda G, Jamroziak K, Reis RM, Garcia-Sanz R, et al. Genetics and molecular epidemiology of multiple myeloma: the rationale for the IMMEnSE consortium (review). Int J Oncol 2012;40(3):625e38. [85] Morgan GJ, Johnson DC, Weinhold N, Goldschmidt H, Landgren O, Lynch HT, et al. Inherited genetic susceptibility to multiple myeloma. Leukemia 2014;28(3):518e24. [86] Wallin A, Larsson SC. Body mass index and risk of multiple myeloma: a metaanalysis of prospective studies. Eur J Cancer (Oxford, Engl: 1990) 2011;47(11):1606e15. [87] Alexander DD, Mink PJ, Adami HO, Cole P, Mandel JS, Oken MM, et al. Multiple myeloma: a review of the epidemiologic literature. Int J Cancer 2007;120(Suppl 12):40e61. [88] Becker N. Epidemiology of multiple myeloma. Recent Results Cancer Res 2011;183:25e35. [89] De Roos AJ, Baris D, Weiss NS, Herrington L. Chapter 47. Multiple myeloma. In: Schottenfeld D, Fraumeni Jr JF, editors. Cancer Epidemiology and Prevention. 3rd ed. New York: Oxford University Press; 2006. p. 919e45. [90] Grulich AE, van Leeuwen MT, Falster MO, Vajdic CM. Incidence of cancers in people with HIV/AIDS compared with immunosuppressed transplant recipients: a meta-analysis. Lancet 2007;370(9581):59e67. [91] McShane CM, Murray LJ, Landgren O, O’Rorke MA, Korde N, Kunzmann AT, et al. Prior autoimmune disease and risk of monoclonal gammopathy of undetermined significance and multiple myeloma: a systematic review. Cancer Epidemiol Biomarkers Prev 2014;23(2):332e42. [92] Brown LM, Gridley G, Check D, Landgren O. Risk of multiple myeloma and monoclonal gammopathy of undetermined significance among white and black male United States veterans with prior autoimmune, infectious, inflammatory, and allergic disorders. Blood 2008;111(7):3388e94. [93] McShane CM, Murray LJ, Engels EA, Landgren O, Anderson LA. Common community-acquired infections and subsequent risk of multiple myeloma: a population-based study. Int J Cancer 2014;134(7):1734e40. [94] Rota M, Porta L, Pelucchi C, Negri E, Bagnardi V, Bellocco R, et al. Alcohol drinking and multiple myeloma riskea systematic review and meta-analysis of the dose-risk relationship. Eur J Cancer Prev 2014;23(2):113e21. [95] Psaltopoulou T, Sergentanis TN, Kanellias N, Kanavidis P, Terpos E, Dimopoulos MA. Tobacco smoking and risk of multiple myeloma: a metaanalysis of 40 observational studies. Int J Cancer 2013;132(10):2413e31. [96] Hsu WL, Preston DL, Soda M, Sugiyama H, Funamoto S, Kodama K, et al. The incidence of leukemia, lymphoma and multiple myeloma among atomic bomb survivors: 1950-2001. Radiat Res 2013;179(3):361e82. [97] Broderick P, Chubb D, Johnson DC, Weinhold N, Forsti A, Lloyd A, et al. Common variation at 3p22.1 and 7p15.3 influences multiple myeloma risk. Nat Genet 2012;44(1):58e61. [98] Chubb D, Weinhold N, Broderick P, Chen B, Johnson DC, Forsti A, et al. Common variation at 3q26.2, 6p21.33, 17p11.2 and 22q13.1 influences multiple myeloma risk. Nat Genet 2013;45(10):1221e5. [99] Fonseca R, Bergsagel PL, Drach J, Shaughnessy J, Gutierrez N, Stewart AK, et al. International Myeloma Working Group molecular classification of multiple myeloma: spotlight review. Leukemia 2009;23(12):2210e21. [100] Boffetta P, Mundt KA, Adami HO, Cole P, Mandel JS. TCDD and cancer: a critical review of epidemiologic studies. Crit Rev Toxicol 2011;41(7):622e36. [101] Chang ET, Boffetta P, Adami HO, Cole P, Mandel JS. A critical review of the epidemiology of Agent Orange/TCDD and prostate cancer. Eur J Epidemiol 2014;29(10):667e723. [102] IARC. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Volume 69. Polychlorinated Dibenzo-para-Dioxins and Polychlorinated Dibenzofurans. Lyon: International Agency for Research on Cancer (IARC); 1997. p. 666. [103] Egger M, Schneider M, Davey Smith G. Spurious precision? Meta-analysis of observational studies. BMJ 1998;316(7125):140e4. [104] Weed DL. Meta-analysis and causal inference: a case study of benzene and non-Hodgkin lymphoma. Ann Epidemiol 2010;20(5):347e55. [105] Shapiro S. Meta-analysis/Shmeta-analysis. Am J Epidemiol 1994;140(9): 771e8. [106] Hill AB. The Environment and Disease: Association or Causation? Proc R Soc Med 1965;58:295e300. [107] Air Force Health Study. Air Force Health Study Comprehensive Report. An Epidemiologic Investigation of Health Effects in Air Force Personnel Following Exposure to Herbicides. February 1984 to March 2005. Prepared

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

[108]

[109]

[110]

[111] [112]

[113]

[114]

[115]

[116] [117]

[118]

[119]

[120] [121]

[122]

[123]

[124] [125] [126]

[127]

[128]

[129]

[130]

[131]

[132] [133] [134]

for the United States Air Force (USAF) by Science Applications International Corporation (SAIC), 2005. Wolfe WH, Michalek JE, Miner JC, Rahe A, Silva J, Thomas WF, et al. Health status of Air Force veterans occupationally exposed to herbicides in Vietnam. I. Physical health. J Am Med Assoc 1990;264(14):1824e31. Pavuk M, Patterson Jr DG, Turner WE. Serum concentrations of TCDD and other dioxin-like compounds in US Air Force veterans of Operation Ranch Hand. Chemosphere 2014;102:18e23. Air Force Health Study. An Epidemiologic Investigation of Health Effects in Air Force Personnel Following Exposure to Herbicides. Volume I. Serum Dioxin Analysis of 1987 Examination Results. McLean, Virginia, and Brooks Air Force Base. Texas: Science Applications International Corporation and Epidemiology Research Division, Armstrong Laboratory; 1991. Ketchum NS, Michalek JE, Burton JE. Serum dioxin and cancer in veterans of Operation Ranch Hand. Am J Epidemiol 1999;149(7):630e9. Michalek JE, Ketchum NS, Akhtar FZ. Postservice mortality of US Air Force veterans occupationally exposed to herbicides in Vietnam: 15-year followup. Am J Epidemiol 1998;148(8):786e92. Pavuk M, Michalek JE, Schecter A, Ketchum NS, Akhtar FZ, Fox KA. Did TCDD exposure or service in Southeast Asia increase the risk of cancer in air force Vietnam veterans who did not spray Agent Orange? J Occup Environ Med 2005;47(4):335e42. Ketchum NS, Akhtar FZ, The Air Force Health Study. An Epidemiologic Investigation of Health Effects in Air Force Personnel Following Exposure to Herbicides. Mortality Update 1996. Brooks Air Force Base. Texas: Air Force Materiel Command; 1996. Ketchum NS, Michalek JE. Postservice mortality of Air Force veterans occupationally exposed to herbicides during the Vietnam War: 20-year follow-up results. Mil Med 2005;170(5):406e13. Akhtar FZ, Garabrant DH, Ketchum NS, Michalek JE. Cancer in US Air Force veterans of the Vietnam War. J Occup Environ Med 2004;46(2):123e36. Kang HK, Dalager NA, Needham LL, Patterson Jr DG, Matanoski GM, Kanchanaraksa S, et al. US Army Chemical Corps Vietnam veterans health study: preliminary results. Chemosphere 2001;43(4-7):943e9. Kang HK, Dalager NA, Needham LL, Patterson Jr DG, Lees PS, Yates K, et al. Health status of Army Chemical Corps Vietnam veterans who sprayed defoliant in Vietnam. Am J Ind Med 2006;49(11):875e84. Centers for Disease Control Veterans Health Studies. Serum 2,3,7,8tetrachlorodibenzo-p-dioxin levels in US Army Vietnam-era veterans. The Centers for Disease Control Veterans Health Studies. J Am Med Assoc 1988;260(9):1249e54. Thomas TL, Kang HK. Mortality and morbidity among Army Chemical Corps Vietnam veterans: a preliminary report. Am J Ind Med 1990;18(6):665e73. Cypel Y, Kang H. Mortality patterns of Army Chemical Corps veterans who were occupationally exposed to herbicides in Vietnam. Ann Epidemiol 2010;20(5):339e46. Selected Cancers Cooperative Study Group. The association of selected cancers with service in the US Military in Vietnam: III. Hodgkin’s disease, nasal cancer, nasopharyngeal cancer, and primary liver cancer. Arch Intern Med 1990;150(12):2495e505. Selected Cancers Cooperative Study Group. The association of selected cancers with service in the US Military in Vietnam I. Non-Hodgkin’s lymphoma. Arch Intern Med 1990;150(12):2473e83. Dalager NA, Kang HK, Burt VL, Weatherbee L. Non-Hodgkin’s lymphoma among Vietnam veterans. J Occup Med 1991;33(7):774e9. Dalager NA, Kang HK, Burt VL, Weatherbee L. Hodgkin’s disease and Vietnam service. Ann Epidemiol 1995;5(5):400e6. Yi SW, Ohrr H, Hong JS, Yi JJ. Agent Orange Exposure and Prevalence of Selfreported Diseases in Korean Vietnam Veterans. J Prev Med Public Health 2013;46(5):213e25. Yi SW, Ohrr H. Agent Orange exposure and cancer incidence in Korean Vietnam veterans: A prospective cohort study. Cancer 2014;120(23): 3699e706. Stellman JM, Stellman SD, Weber T, Tomasallo C, Stellman AB, Christian Jr R. A geographic information system for characterizing exposure to Agent Orange and other herbicides in Vietnam. Environ Health Perspect 2003; 111(3):321e8. Yi SW, Ohrr H, Won JU, Song JS, Hong JS. Serum 2,3,7,8-Tetrachlorodibenzop-dioxin Levels and Their Association With Age, Body Mass Index, Smoking, Military Record-based Variables, and Estimated Exposure to Agent Orange in Korean Vietnam Veterans. J Prev Med Public Health 2013;46(5):226e36. Chen Q, Garabrant DH, Hedgeman E, Little RJ, Elliott MR, Gillespie B, et al. Estimation of background serum 2,3,7,8-TCDD concentrations by using quantile regression in the UMDES and NHANES populations. Epidemiology (Cambridge, Mass) 2010;21(Suppl 4):S51e7. Ginevan ME, Watkins DK, Ross JH, O’Boyle RA. Assessing exposure to allied ground troops in the Vietnam War: a quantitative evaluation of the Stellman Exposure Opportunity Index model. Chemosphere 2009;75(11):1512e8. Young AL, Cecil Sr PF. Agent Orange exposure and attributed health effects in Vietnam veterans. Mil Med 2011;176(7 Suppl):29e34. Shepard CW, Simard EP, Finelli L, Fiore AE, Bell BP. Hepatitis B virus infection: epidemiology and vaccination. Epidemiol Rev 2006;28:112e25. Dominitz JA, Boyko EJ, Koepsell TD, Heagerty PJ, Maynard C, Sporleder JL, et al. Elevated prevalence of hepatitis C infection in users of United States veterans medical centers. Hepatology (Baltimore, Md) 2005;41(1):88e96.

291

[135] Brau N, Bini EJ, Shahidi A, Aytaman A, Xiao P, Stancic S, et al. Prevalence of hepatitis C and coinfection with HIV among United States veterans in the New York City metropolitan area. Am J Gastroenterol 2002;97(8):2071e8. [136] Briggs ME, Baker C, Hall R, Gaziano JM, Gagnon D, Bzowej N, et al. Prevalence and risk factors for hepatitis C virus infection at an urban Veterans Administration medical center. Hepatology (Baltimore, Md) 2001;34(6):1200e5. [137] Owens DK, Sundaram V, Lazzeroni LC, Douglass LR, Sanders GD, Taylor K, et al. Prevalence of HIV infection among inpatients and outpatients in Department of Veterans Affairs health care systems: implications for screening programs for HIV. Am J Public Health 2007;97(12):2173e8. [138] Centers for Disease Control and Prevention (CDC). HIV prevalence estimateseUnited States, 2006. MMWR Morbidity Mortality Weekly Rep 2008; 57(39):1073e6. [139] Wilson EJ, Horsley KW, van der Hoek R. Australian National Service Vietnam Veterans: Mortality and Cancer Incidence 2005. Canberra, Australia: Australian Government, Department of Veterans’ Affairs, Australian Institute of Health and Welfare; 2005. [140] Commonwealth Department of Veterans’ Affairs. Morbidity of Vietnam Veterans: A Study of the Health of Australia’s Vietnam Veteran Community: Volume 2-Female Vietnam Veterans. Survey and Community Comparison Outcomes. Canberra, Australia: Australian Government, Department of Veterans’ Affairs; 1998. [141] O’Brien TR, Decoufle P, Boyle CA. Non-Hodgkin’s lymphoma in a cohort of Vietnam veterans. Am J Public Health 1991;81(6):758e60. [142] Boehmer TK, Flanders WD, McGeehin MA, Boyle C, Barrett DH. Postservice mortality in Vietnam veterans: 30-year follow-up. Arch Intern Med 2004; 164(17):1908e16. [143] Thomas TL, Kang HK, Dalager NA. Mortality among women Vietnam veterans, 1973-1987. Am J Epidemiol 1991;134(9):973e80. [144] Watanabe KK, Kang HK. Military service in Vietnam and the risk of death from trauma and selected cancers. Ann Epidemiol 1995;5(5):407e12. [145] McBride D, Cox B, Broughton J, Tong D. The mortality and cancer experience of New Zealand Vietnam war veterans: a cohort study. BMJ Open 2013;3(9): e003379. [146] Yi SW. Cancer incidence in Korean Vietnam Veterans during 1992-2003: The Korean Veterans Health Study. J Prev Med Public Health 2013;46(6): 309e18. [147] Commonwealth Department of Veterans’ Affairs. Morbidity of Vietnam Veterans: A Study of the Health of Australia’s Vietnam Veteran Community: Volume 1-Male Vietnam Veterans. Survey and Community Comparison Outcomes. Canberra, Australia: Australian Government, Department of Veterans’ Affairs; 1998. [148] Australian Institute of Health and Welfare. Morbidity of Vietnam Veterans. A Study of the Health of Australia’s Vietnam Veteran Community. Volume 3 Validation Study. Canberra, Australia: Australian Institute of Health and Welfare; 1999. [149] Wilson EJ, Horsley KW, van der Hoek R. Cancer Incidence in Australian Vietnam Veterans Study 2005. Canberra, Australia: Australian Government, Department of Veterans’ Affairs, Australian Institute of Health and Welfare; 2005. [150] Wilson EJ, Horsley KW, van der Hoek R. The Third Australian Vietnam Veterans Mortality Study 2005. Canberra, Australia: Australian Government, Department of Veterans’ Affairs, Australian Institute of Health and Welfare; 2005. [151] Anderson HA, Hanrahan LP, Jensen M, Laurin D, Yick WY, Wiegman P. Wisconsin Vietnam Veteran Mortality Study. Madison, WI: State of Wisconsin, Department of Health and Social Services, Division of Health, Section of Environmental and Chronic Disease Epidemiology; 1986. [152] Hall W. The logic of a controversy: the case of Agent Orange in Australia. Social Sci Med 1989;29(4):537e44. [153] Rothman KJ, Greenland S, Lash TL. Chapter 6. Types of epidemiologic studies. In: Rothman KJ, Greenland S, Lash TL, editors. Modern Epidemiology. 3rd ed. Philadelphia: Lippincott Williams & Wilkins; 2008. p. 97e9. [154] Lawrence CE, Reilly AA, Quickenton P, Greenwald P, Page WF, Kuntz AJ. Mortality patterns of New York State Vietnam Veterans. Am J Public Health 1985;75(3):277e9. [155] Holmes AP, Bailey C, Baron RC, Bosanac E, Brough J, Conroy C, et al. West Virginia Vietnam-Era Veterans Mortality Study. Preliminary Report. Charleston, WV: West Virginia Health Department; 1986. [156] Clapp RW, Cupples LA, Colton T, Ozonoff DM. Cancer surveillance of Veterans in Massachusetts, USA, 1982-1988. Int J Epidemiol 1991;20(1):7e12. [157] Clapp RW. Update of cancer surveillance of veterans in Massachusetts, USA. Int J Epidemiol 1997;26(3):679e81. [158] Kogan MD, Clapp R. Mortality Among Vietnam Veterans in Massachusetts, 1972-1983. Boston, MA: Massachusetts Office of Commissioner of Veterans Services, Agent Orange Program, Massachusetts Department of Public Health, Division of Health Statistics and Research; 1985. [159] Visintainer PF, Barone M, McGee H, Peterson EL. Proportionate mortality study of Vietnam-era veterans of Michigan. J Occup Environ Med 1995;37(4):423e8. [160] Watanabe KK, Kang HK. Mortality patterns among Vietnam veterans: a 24-year retrospective analysis. J Occup Environ Med 1996;38(3):272e8. [161] Bullman TA, Kang HK, Watanabe KK. Proportionate mortality among US Army Vietnam veterans who served in military region I. Am J Epidemiol 1990;132(4):670e4.

292

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

[162] Saracci R, Kogevinas M, Bertazzi PA, Bueno de Mesquita BH, Coggon D, Green LM, et al. Cancer mortality in workers exposed to chlorophenoxy herbicides and chlorophenols. Lancet 1991;338(8774):1027e32. [163] Fingerhut MA, Halperin WE, Marlow DA, Piacitelli LA, Honchar PA, Sweeney MH, et al. Cancer mortality in workers exposed to 2,3,7,8tetrachlorodibenzo-p-dioxin. New Engl J Med 1991;324(4):212e8. [164] Kogevinas M, Becher H, Benn T, Bertazzi PA, Boffetta P, Bueno-deMesquita HB, et al. Cancer mortality in workers exposed to phenoxy herbicides, chlorophenols, and dioxins. An expanded and updated international cohort study. Am J Epidemiol 1997;145(12):1061e75. [165] Steenland K, Piacitelli L, Deddens J, Fingerhut M, Chang LI. Cancer, heart disease, and diabetes in workers exposed to 2,3,7,8-tetrachlorodibenzo-pdioxin. J Natl Cancer Inst 1999;91(9):779e86. [166] Boers D, Portengen L, Bueno-de-Mesquita HB, Heederik D, Vermeulen R. Cause-specific mortality of Dutch chlorophenoxy herbicide manufacturing workers. Occup Environ Med 2010;67(1):24e31. [167] Boers D, Portengen L, Turner WE, Bueno-de-Mesquita HB, Heederik D, Vermeulen R. Plasma dioxin levels and cause-specific mortality in an occupational cohort of workers exposed to chlorophenoxy herbicides, chlorophenols and contaminants. Occup Environ Med 2012;69(2):113e8. [168] Collins JJ, Bodner K, Aylward LL, Wilken M, Bodnar CM. Mortality rates among trichlorophenol workers with exposure to 2,3,7,8-tetrachlorodibenzo-pdioxin. Am J Epidemiol 2009;170(4):501e6. [169] Collins JJ, Bodner K, Aylward LL, Wilken M, Swaen G, Budinsky R, et al. Mortality rates among workers exposed to dioxins in the manufacture of pentachlorophenol. J Occup Environ Med 2009;51(10):1212e9. [170] Manuwald U, Velasco Garrido M, Berger J, Manz A, Baur X. Mortality study of chemical workers exposed to dioxins: follow-up 23 years after chemical plant closure. Occup Environ Med 2012;69(9):636e42. [171] McBride DI, Collins JJ, Humphry NF, Herbison P, Bodner KM, Aylward LL, et al. Mortality in workers exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin at a trichlorophenol plant in New Zealand. J Occup Environ Med 2009;51(9): 1049e56. [172] Ruder AM, Yiin JH. Mortality of US pentachlorophenol production workers through 2005. Chemosphere 2011;83(6):851e61. [173] ’t Mannetje A, McLean D, Cheng S, Boffetta P, Colin D, Pearce N. Mortality in New Zealand workers exposed to phenoxy herbicides and dioxins. Occup Environ Med 2005;62(1):34e40. [174] Ott MG, Zober A. Cause specific mortality and cancer incidence among employees exposed to 2,3,7,8-TCDD after a 1953 reactor accident. Occup Environ Med 1996;53(9):606e12. [175] Neuberger M, Landvoigt W, Derntl F. Blood levels of 2,3,7,8tetrachlorodibenzo-p-dioxin in chemical workers after chloracne and in comparison groups. Int Arch Occup Environ Health 1991;63(5):325e7. [176] Flesch-Janys D, Berger J, Gurn P, Manz A, Nagel S, Waltsgott H, et al. Exposure to polychlorinated dioxins and furans (PCDD/F) and mortality in a cohort of workers from a herbicide-producing plant in Hamburg, Federal Republic of Germany. Am J Epidemiol 1995;142(11):1165e75. [177] Manz A, Berger J, Dwyer JH, Flesch-Janys D, Nagel S, Waltsgott H. Cancer mortality among workers in chemical plant contaminated with dioxin. Lancet 1991;338(8773):959e64. [178] Smith AH, Patterson Jr DG, Warner ML, MacKenzie R, Needham LL. Serum 2,3,7,8-tetrachlorodibenzo-p-dioxin levels of New Zealand pesticide applicators and their implication for cancer hypotheses. J Natl Cancer Inst 1992;84(2):104e8. [179] Littorin M, Hansson M, Rappe C, Kogevinas M. Dioxins in blood from Swedish phenoxy herbicide workers. Lancet 1994;344(8922):611e2. [180] Kogevinas M, Kauppinen T, Winkelmann R, Becher H, Bertazzi PA, Bueno-deMesquita HB, et al. Soft tissue sarcoma and non-Hodgkin’s lymphoma in workers exposed to phenoxy herbicides, chlorophenols, and dioxins: two nested case-control studies. Epidemiology (Cambridge, Mass) 1995;6(4): 396e402. [181] Jurek AM, Greenland S, Maldonado G, Church TR. Proper interpretation of non-differential misclassification effects: expectations vs observations. Int J Epidemiol 2005;34(3):680e7. [182] Consonni D, Pesatori AC, Zocchetti C, Sindaco R, D’Oro LC, Rubagotti M, et al. Mortality in a population exposed to dioxin after the Seveso, Italy, accident in 1976: 25 years of follow-up. Am J Epidemiol 2008;167(7):847e58. [183] Pesatori AC, Consonni D, Rubagotti M, Grillo P, Bertazzi PA. Cancer incidence in the population exposed to dioxin after the “Seveso accident”: twenty years of follow-up. Environ Health 2009;8:39. [184] Warner M, Mocarelli P, Samuels S, Needham L, Brambilla P, Eskenazi B. Dioxin exposure and cancer risk in the Seveso Women’s Health Study. Environ Health Perspect 2011;119(12):1700e5. [185] Hardell L, Eriksson M, Athlin L, Hansson M, Rappe C. Adipose-tissue concentrations of dioxins and dibenzofurans in potentially exposed patients with malignant-lymphoma or sarcoma. Oncol Rep 1995;2(5):749e53. [186] Hardell L, Fredrikson M, Eriksson M, Hansson M, Rappe C. Adipose tissue concentrations of dioxins and dibenzofurans in patients with malignant lymphoproliferative diseases and in patients without a malignant disease. Eur J Cancer Prev 1995;4(3):225e9.

[187] Hardell L, Eriksson M, Lindstrom G, Van Bavel B, Linde A, Carlberg M, et al. Case-control study on concentrations of organohalogen compounds and titers of antibodies to Epstein-Barr virus antigens in the etiology of nonHodgkin lymphoma. Leuk Lymphoma 2001;42(4):619e29. [188] Hardell L, Vanbavel B, Lindstrom G, Fredrikson M, Hagberg H, Liljegren G, et al. Higher concentrations of specific polychlorinated biphenyl congeners in adipose tissue from non-Hodgkin’s lymphoma patients compared with controls without a malignant disease. Int J Oncol 1996;9(4):603e8. [189] Kim MJ, Marchand P, Henegar C, Antignac JP, Alili R, Poitou C, et al. Fate and complex pathogenic effects of dioxins and polychlorinated biphenyls in obese subjects before and after drastic weight loss. Environ Health Perspect 2011;119(3):377e83. [190] Kramárová E, Kogevinas M, Anh CT, Cau HD, Dai LC, Stellman SD, et al. Exposure to Agent Orange and occurrence of soft-tissue sarcomas or nonhodgkin lymphomas: An ongoing study in Vietnam. Environ Health Perspect 1998;106(Suppl 2):671e8. [191] Kramarova E, Parkin DM, Cung TA, Viet AT, Vu NP, Le VX, et al. Case-control study of soft-tissue sarcomas and non-Hodgkin lymphomas in relation to exposure to herbicides in Vietnam. Proceedings of the Vietnam-United States Scientific Conference on Human Health and Environmental Effects of Agent Orange/Dioxins, 3-6 March 2002, Hanoi, Vietnam Part 2: Human Health Effects. 2004:726e733. [192] Stellman SD, Stellman JM. Estimation of exposure to Agent Orange and other defoliants among American troops in Vietnam: a methodological approach. Am J Ind Med 1986;9(4):305e21. [193] De Roos AJ, Hartge P, Lubin JH, Colt JS, Davis S, Cerhan JR, et al. Persistent organochlorine chemicals in plasma and risk of non-Hodgkin’s lymphoma. Cancer Res 2005;65(23):11214e26. [194] Morgenstern H. Ecologic studies in epidemiology: concepts, principles, and methods. Annu Rev Public Health 1995;16:61e81. [195] Mannes T, Emmett K, Willmore A, Churches T, Sheppeard V, Kaldor J. Effect of a chemical manufacturing plant on community cancer rates. BMC Public Health 2005;5:34. [196] Read D, Wright C, Weinstein P, Borman B. Cancer incidence and mortality in a New Zealand community potentially exposed to 2,3,7,8-tetrachlorodibenzop-dioxin from 2,4,5-trichlorophenoxyacetic acid manufacture. Aust N Z J Public Health 2007;31(1):13e8. [197] Revich B, Aksel E, Ushakova T, Ivanova I, Zhuchenko N, Klyuev N, et al. Dioxin exposure and public health in Chapaevsk, Russia. Chemosphere 2001;43(4-7):951e66. [198] Cogliano VJ, Baan R, Straif K, Grosse Y, Lauby-Secretan B, El Ghissassi F, et al. Preventable exposures associated with human cancers. J Natl Cancer Inst 2011;103(24):1827e39. [199] Cole P, Trichopoulos D, Pastides H, Starr T, Mandel JS. Dioxin and cancer: a critical review. Regul Toxicol Pharmacol 2003;38(3):378e88. [200] Clarke CA, Morton LM, Lynch C, Pfeiffer RM, Hall EC, Gibson TM, et al. Risk of lymphoma subtypes after solid organ transplantation in the United States. Br J Cancer 2013;109(1):280e8. [201] Sherr DH, Monti S. The role of the aryl hydrocarbon receptor in normal and malignant B cell development. Semin Immunopathol 2013;35(6):705e16. [202] Stange J, Veldhoen M. The aryl hydrocarbon receptor in innate T cell immunity. Semin Immunopathol 2013;35(6):645e55. [203] Hao N, Whitelaw ML. The emerging roles of AhR in physiology and immunity. Biochem Pharmacol 2013;86(5):561e70. [204] Nguyen NT, Hanieh H, Nakahama T, Kishimoto T. The roles of aryl hydrocarbon receptor in immune responses. Int Immunol 2013;25(6): 335e43. [205] IARC. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Volume 69. Polychlorinated Dibenzo-para-Dioxins and Polychlorinated Dibenzofurans. Geneva: International Agency for Research on Cancer (IARC); 1997. [206] Howlader N, Noone AM, Krapcho M, Garshell J, Miller D, Altekruse SF, et al, editors. SEER Cancer Statistics Review, 1975-2011, National Cancer Institute. Bethesda, MD, http://seer.cancer.gov/csr/1975_2011/, based on November 2013 SEER data submission 2014. [207] IARC. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Volume 41. Some Halogenated Hydrocarbons and Pesticide Exposures. Geneva: International Agency for Research on Cancer (IARC); 1986. [208] Parkin DM, Whelan SL, Ferlay J, Raymond L, Young J, editors. Cancer Incidence in Five Continents, Vol. VII. Lyon: IARC; 1997. IARC Scientific Publications No. 143. [209] Parkin DM, Whelan SL, Ferlay J, Teppo L, Thomas DB, editors. Cancer Incidence in Five Continents, Vol. VIII. Lyon: IARC; 2002. IARC Scientific Publications No. 155. [210] Morton LM, Slager SL, Cerhan JR, Wang SS, Vajdic CM, Skibola CF, et al. Etiologic heterogeneity among NHL subtypes: The InterLymph NHL subtypes project. J Natl Cancer Inst Monogr 2014;2014(48):130e44. [211] Crane PJ, Barnard DL, Horsley KD, Adena MA. Mortality of Vietnam Veterans. The Veteran Cohort Study. A Report of the 1996 Retrospective Cohort Study of Australian Vietnam Veterans. Canberra, Australia: Australian Government, Department of Veterans’ Affairs; 1997.

Appendix Supplementary Table 1 Characteristics of epidemiologic studies of Agent Orange/TCDD exposure and lymphoid malignancies Author, publication year

Study subjects

Studies of veterans with estimated Agent Orange/TCDD exposure Air Force Health Study Ketchum and Akhtar, 1996 [114]; 1261 Male U.S. Air Force veterans of Michalek et al., 1998 [112] Operation Ranch Hand and 19,101 comparison male U.S. Air Force veterans involved in other C-130 aircraft missions in Southeast Asia In [112], 31 comparison subjects with missing service dates and one erroneously classified comparison subject were excluded, and 11 other comparison subjects were added

Period of exposure and follow-up

Definition of outcome

Definition of exposure

Active service between August 1961 and May 1972; spray operations between January 1962 and October 1971 Mortality follow-up from the first qualifying tour of duty through December 31, 1993

Reticulosarcoma, reticulolymphosarcoma, other lymphoma, CLL, HL, and MM mortality ascertained using U.S. Air Force, VA Beneficiary Identification Records Locator Subsystem, and Internal Revenue Service records; underlying cause of death on death certificate (obtained for 99.8% of Ranch Hands and 99.9% of comparison subjects) classified according to National Center for Health Statistics decision tables

Service in Operation Ranch Hand Median (range) serum TCDD levels in 1987, ppt [108] Ranch Hands (n ¼ 888): 12.4 (0.0e617.7) - Flying officers (pilot): 7.3 (0.0e42.6) - Flying officers (navigator): 9.3 (1.1e35.9) - Nonflying officers: 6.6 (3.1e24.9) - Flying enlisted personnel: 17.2 (0.0e195.5) - Nonflying enlisted personnel: 23.6 (0.0e617.7)

AFHS, 2005 [107]

1197 Male U.S. Air Force veterans of Operation Ranch Hand and 1,855 randomly selected comparison male U.S. Air Force veterans involved in other C-130 aircraft missions in Southeast Asia, matched on date of birth, race, rank, and military occupation, with available medical records

Active service between August 1961 and May 1972; spray operations between January 1962 and October 1971 Cancer morbidity follow-up from 1982 through 2002

NHL, CLL, HL, and MM morbidity and mortality ascertained from periodic physical examinations, questionnaires, and telephone interviews in 1982, 1985, 1987, 1992, 1997, and 2002, medical records (to validate reported diagnoses and identify unreported diagnoses), and death certificates; examinations conducted and records reviewed by personnel blinded to exposure status

Ketchum and Michalek, 2005 [115]

1262 Male U.S. Air Force veterans of Operation Ranch Hand and 19,078 comparison male U.S. Air Force veterans involved in other C-130 aircraft missions in Southeast Asia After [112], one comparison subject was reclassified as a Ranch Hand and another was reclassified as ineligible

Active service between August 1961 and May 1972; spray operations between January 1962 and October 1971 Mortality follow-up from the first qualifying tour of duty through December 31, 1999

Malignant lymphoma, NHL, HL, lymphoreticulosarcoma, and MM mortality; see Ketchum and Akhtar (1996)

894 Male veterans who served in at least one U.S. Army Chemical Corps unit assigned to Vietnam between 1966 and 1971 (mean age at end of service in Vietnam ¼ 25 y), identified from morning reports by the U.S. Army and Joint Services Environmental Support Group Comparisons with U.S. male national

Assigned to Vietnam between 1966 and 1971 Mortality follow-up from date last served in Vietnam through December 31, 1987; mean follow-up ¼ 18 y Morbidity follow-up in Agent Orange Registry from 1978 to 1988, VA Patient Treatment File from 1970 to 1986

NHL and HL mortality ascertained from VA Beneficiary Identification Records Locator Subsystem, Social Security Administration, U.S. Internal Revenue Service, National Death Index, and military records; underlying cause of death coded from death certificate for all subjects known to be deceased

Army Chemical Corps study Thomas and Kang, 1990 [120]

- Flying officers (navigator): 4.5 (2.2e7.9) - Nonflying officers: 4.3 (0.0e6.0) - Flying enlisted personnel: 4.0 (0.0e12.7) - Nonflying enlisted personnel: 3.9 (0.0e54.8) Service in Operation Ranch Hand TCDD levels measured on a lipid weight basis in serum in 1987, 1992, 1997, or 2002 in 1027 (86%) of 1197 Ranch Hands and 1524 (82%) of 1855 comparison subjects; levels extrapolated to end of qualifying tour of duty in Vietnam assuming constant half-life of 7.6 y; categorized as “background” (10 ppt), “low” (>10 to 118.5 ppt), or “high” (>118.5 ppt) Service in Operation Ranch Hand Median (95% CI) serum TCDD levels in 2002, pg/g lipid [108] Ranch Hands (n ¼ 777): 5.0 (4.6e5.5) Comparison subjects (n ¼ 1173): 2.2 (2.1e2.2)

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Comparison subjects (n ¼ 856): 4.2 (0.0e54.8) - Flying officers (pilot): 4.7 (0.0e18.5)

Service in Army Chemical Corps unit assigned to Vietnam between 1966 and 1971

292.e1

(continued on next page)

Author, publication year

Study subjects

Period of exposure and follow-up

mortality rates standardized by age, race, and calendar year No comparisons for morbidity data

Cypel and Kang, 2010 [121]

Active service between July 1, 1965, and March 28, 1973 Mortality follow-up from date of the last military service or March 28, 1973 (whichever was earlier), through December 31, 2005; mean follow-up in both cohorts z32 y

1157 Men with incident, pathologically confirmed NHL (mean age in 1968 ¼ 29 y) and 310 with HL (mean age in 1968 ¼ 25 y) diagnosed December 1, 1984, to November 30, 1988, who lived in the metropolitan areas of Atlanta, Georgia; Detroit, Michigan; San Francisco, California; Seattle, Washington; and Miami, Florida; and in the states of Connecticut, Iowa, and Kansas (diagnosed on or after October 1, 1985, in Seattle and Detroit), born in 1929e1953; 88.1% participation rate; median of 103 days between diagnosis and interview (10% proxy for deceased) 1776 Controls identified by random-digit dialling, frequency matched to lymphoma cases by geographic area and year of birth (mean age in 1968 ¼ 27 y); median of 46 d between receipt of control information and interview; 90% completion rate for eligibility screening among 15,700 households; 91% identification of name and address among households with an eligible man; 83.1% completion rate among selected men; deceased controls (excluding deaths from six selected cancers, homicide, or suicide) chosen from death certificate files and matched to each deceased NHL case on geography,

Military service between 1964 and 1972 Cases diagnosed 1984e1988

Definition of exposure

NHL and HL morbidity ascertained from linkage to the VA Agent Orange Registry of Vietnam veterans who volunteered for a special physical examination at a VA medical center since 1978 (168 matches with study cohort), and Patient Treatment File of inpatient discharges at all 172 U.S. VA medical centers since 1970 (136 matches with study cohort; 257 total matches) Lymphoid cancer mortality (calculated from lymphopoietic cancer minus leukemia) ascertained from VA Beneficiary Identification Records Locator Subsystem, Social Security Administration, and National Death Index; underlying cause of death coded from death certificates obtained for decedents through 1991 and from National Death Index Plus thereafter Cause of death ascertained for 98% of Vietnam veterans, 95% of non-Vietnam veterans

Service in Army Chemical Corps unit assigned to Vietnam between 1965 and 1973

Incident, pathologically confirmed NHL or HL; microscopic slides or tissue blocks obtained from pathology department of diagnosis for 97% of lymphoma cases interviewed, confirmed as lymphoma by an expert panel of three pathologists for 93% cases; 129 of 136 unconfirmed cases were considered to lack an adequate specimen; NHL cases classified according to Working Formulation

Military service in Vietnam assessed by telephone interview; subjects who reported active U.S. military service were asked whether they were “stationed in Vietnam or off the coast of Vietnam,” classified as exposed if yes; 40 others who were in or off the coast of Vietnam in the military but not stationed there, mostly only briefly, were included as exposed in sensitivity analyses. 96 of 115 possible NHL cases and 98 of 133 controls who reported military service in Vietnam consented to review of military records; records located for 86% of NHL cases and 87% of controls; Vietnam service verified for 86% of NHL cases and 92% controls with records Vietnam veterans reported dates, branch of service, and rank for each tour of duty, and unit, location in Vietnam, job duties, and self-perceived exposure to herbicides. Units and branches classified by U.S. Army and Joint Services Environmental Support Group as combat, combat support, or support, based on interview information; Navy assignments classified as blue-water, brown-water, or on-shore; other assignments classified as I, II, III, or IV Corps Occupational exposure to phenoxy

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Selected Cancers Study Selected Cancers Cooperative Study Group, 1990 [123, 122]

2872 Male U.S. Army veterans with 1 assignment to the Chemical Corps in Vietnam (mean age at start of follow-up ¼ 24.0 y, end of follow-up ¼ 58.8 y; 93,318 person-years), identified from morning reports, Defense Manpower Data Center, and Army Chemical School rosters; and 2737 comparison male veterans with the last U.S. Army discharge from Chemical Corps, but no service in Southeast Asia (mean age at start of follow-up ¼ 23.0 y, end of follow-up ¼ 57.1 y; 89,850 personyears) External comparisons with U.S. male national mortality rates standardized by age, race, and calendar year

Definition of outcome

292.e2

Supplementary Table 1 (continued )

herbicides, chlorophenols, or TCDD outside Vietnam assessed based on questions about specific herbicides used (refined with help of toxicologists) and job responsibilities at incinerators, leather-tanning plants, saw mills, meatpacking plants, and jobs involving contact with electrical transformers, cutting oils, hexachlorophene, and specific wood preservatives

year of birth, race, and time from death to interview Excluded men with unknown military or Vietnam service status, military station in or off Vietnam coast but not in Vietnam, AIDS or AIDS-related illness, or none U.S.-resident status before 1969

Confirmed NHL, identified from the VA Patient Treatment File, with pathology reports from VA hospitals with 10 lymphoma cases reviewed for confirmation by a pathologist blinded to Vietnam service status

Military service in Vietnam between 1965 and 1973, ascertained from National Personnel Records Center, including branch and unit of service, dates and rank at time of entry and discharge, and military occupational specialty codes Indirect indicators of potential Agent Orange exposure: Army or Marine Corps service (vs. Air Force or Navy), combat duty (vs. noncombat), longer duration of service, service during peak spraying years of 1967e1969, and service in III Corps (vs. I, II, or IV Corps)

Discharge diagnosis between 1975 and 1985 (median ¼ 1978 for HL cases, 1979 for controls) See [124] for exposure period

Confirmed HL; see [124] for methods

See [124]

Exposure during the Vietnam War Health survey in July 2004

Prevalent NHL, HL, or MM, self-reported by mailed questionnaire

Self-reported perceived exposure to Agent Orange in Vietnam based on six yes/no questions on spraying, handling spray equipment, presence during spraying, contact on skin or clothing, walking through sprayed area, or having (continued on next page)

292.e3

Military service between July 1965 and March 1973 Discharge diagnosis between 1969 and 1985 (median ¼ 1981 for NHL cases, 1982 for controls)

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

U.S. Department of Veterans Affairs hospitals study Dalager et al., 1991 [124] Male Vietnam-era veterans born 1937e1954 and treated at a VA hospital with 10 lymphoma cases in 1969e1985; 201 cases with NHL (median age at discharge ¼ 34 y) and 358 controls with another diagnosis and no lymphoma history (median age at discharge ¼ 34 y) matched 2:1 on hospital, discharge year of first NHL hospitalization, and birth date, selected from the VA Patient Treatment File, a computerized hospital discharge abstract system Of 450 potential cases and 900 controls, military service records were located and abstracted for 90.7% of cases and 88.3% of controls; 15 cases and 18 controls were further excluded due to military service in Southeast Asia but not Vietnam, 50 cases and 80 cases were excluded because of military service outside July 1965 to March 1973, and 142 cases and 339 controls were excluded for the first VA hospitalization within 5 years of discharge Dalager et al., 1995 [125] 283 Male Vietnam-era veterans with HL and 404 without HL (median age in 1968 ¼ 20 y in both groups) treated at 1 of 82 VA hospitals with 10 lymphoma cases Of 770 potential cases and 1540 controls, military service records were located and abstracted for 93.0% of cases and 88.6% of controls; 15 cases and 29 controls were further excluded because of military service in Southeast Asia but not Vietnam, 71 cases and 116 controls were excluded because of military service outside July 1965 to March 1973, and 347 cases and 815 controls were excluded for the first VA hospitalization before 1975 See [124] for methods Korean veterans study Yi et al., 2013 [126] 114,562 Korean veterans of the Vietnam War identified by the Ministry of National Defense and Ministry of Government Administration and Home Affairs as of June 2004, who completed a written, mailed survey

Author, publication year

Study subjects

Period of exposure and follow-up

Incident NHL, HL, or MM, ascertained from the Korea National Cancer Incidence Database; 98% complete follow-up

NHL identified from death certificates (underlying and contributing causes of death; n ¼ 20 deaths) and cancer registration records in the six countries with cancer registration schemes (Austria, Denmark, Finland, New Zealand, Sweden, and United Kingdom; n ¼ 12 cases); evidence of histologic diagnosis available for all but two cases

Definition of exposure been exposed in any other way; categorized into high (“yes” to either of the first two questions), moderate (“yes” to either of the second two questions), low/none (“yes” to either of the last two questions or “no”/“don’t know” to all) Exposure opportunity index model E4 developed by [128] based on proximity of military unit to sprayed areas; battalion/ company-level units identified in selfreported surveys, division/brigade-level units identified from Ministry of Defense military records. Military coordinate information sent to Stellman team for calculating E4 scores, log-transformed for use as Agent Orange exposure index, categorized into low versus high or none, low, moderate, and high Correlations of log-transformed serum TCDD measured in 2007 in 102 Korean Vietnam veterans (mean ¼ 1.2 ppt, median ¼ 0.9 ppt; n ¼ 2 with TCDD >10 ppt): r2 ¼ 0.07 with division/ brigade-level exposure, r2 ¼ 0.16 with company/battalion-level exposure, and r2 ¼ 0.13 with company/battalion-level exposure [126] Exposure opportunity index model E4 developed by [128] based on proximity of division/brigade-level unit to sprayed areas; see [126]

Exposure to TCDD and other chemicals or mixtures assessed by three industrial hygienists blinded to caseecontrol status, using individual job records, detailed company exposure questionnaires, and company reports; cumulative exposure score calculated for each subject and chemical based on estimated level of exposure and duration of exposure in years, excluding exposures <5 y before the date of selection into study Cumulative exposure <0.005 (<1 d) classified as nonexposed; 0.005 classified as ever-exposed, including low exposure (0.005  exposure < 1; generally <1 y), medium exposure (1  exposure < 10), and high exposure (exposure 10). Also examined cumulative exposure without 5-y lag and

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Exposure during the Vietnam War 180,251 Korean veterans of the Vietnam War identified by the Ministry of National Cancer incidence between 1992 and 2003 Defense and Ministry of Government Administration and Home Affairs as of June 2004, excluding those who were deceased, had emigrated, had unknown residency status, had cancer before 1992, or had indeterminate exposure data (7646 subjects; 4%) Studies of manufacturers and sprayers of herbicides potentially contaminated with TCDD Not reported; see [164], which Kogevinas et al., 1995 [180] 32 NHL cases or deaths and 158 controls additionally included U.S. cohorts matched on age, sex, and country of residence, nested within a cohort of 21,183 male and female workers exposed to phenoxy herbicides and chlorophenols in manufacture or spraying, including 24 cohorts from 11 countries (Austria, Australia, Canada, Denmark, Finland, Germany, Italy, Netherlands, New Zealand, Sweden, and United Kingdom). As of 1991, the cohort included 13,898 exposed workers, 3951 nonexposed workers, and 541 with unknown exposure because of incomplete job history information Yi et al., 2014 [127]

Definition of outcome

292.e4

Supplementary Table 1 (continued )

NHL mortality from death certificates obtained through health and safety administration NHL incidence from medical and hospital records and questionnaires or interviews completed by living cohort members, with physician follow-up for diagnostic confirmation NHL, HL, and MM mortality from death certificates obtained from national mortality registries in 10 countries, place of residence (Italy) or contact to companies, insurance, physicians, and family members (Germany)

based on ranking of subjects within each matched set of 6 (1 ¼ lowest, 6 ¼ highest), with similar results (not reported) Ever-employment in activities related to the TCDD-releasing accident Exposure derived from serum TCDD levels: low (<0.1 mg/kg body weight), medium (0.1e0.99 mg/kg body weight), high (1 mg/kg body weight) 113 Workers with chloracne

243 Male workers who were involved in clean-up, repair, maintenance, or demolition activities at a German plant where an accident in 1953 led to release of TCDD; serum TCDD level available for all cohort members

Employment during 1953e1969 in accident-related activities; follow-up 1953e1992

Kogevinas et al., 1997 [164]

21,863 Male and female workers exposed to phenoxy herbicides and chlorophenols in manufacture or spraying; pooled analysis of 36 cohorts (32 manufacture, 4 spraying) from 12 countries (Austria, Australia, Canada, Denmark, Finland, Germany, Italy, Netherlands, New Zealand, Sweden, United Kingdom, and United States)

Employment and cancer mortality follow-up period varied by cohort and country; overall, they extended between 1939 and 1992

Steenland et al., 1999 [165]

5,132 Male workers employed in 12 U.S. plants manufacturing herbicides contaminated with TCDD, including a subcohort of 608 workers (from all plants) with chloracne and no exposure to pentachlorophenol 1,025 Male and female phenoxy herbicide production workers and 703 sprayers in New Zealand

Ever-employment during 1942e1984; follow-up 1942e1993

NHL, HL, and MM mortality from death certificates obtained from National Death Index, Social Security Administration, and Internal Revenue Service

Employment 1969e1984 (production workers; 813 exposed); registration 1973e1984 (sprayers; 699 exposed); follow-up 1969e2000 Employment during 1937e1980; period of follow-up unspecified

NHL, HL, and MM and other immunoproliferative neoplasm mortality from national registry

Employment in herbicide production or spraying

NHL, HL, and other lymphopoietic cancer (excluding leukemia and aleukemia) mortality from death certificates

Ever-employment in PCP manufacture Estimated cumulative TEQ exposure estimated from serum levels of five dioxins and occupational history: low (<0.7 ppt-years), medium (0.7e3.99 ppt-years), high (>4.0 ppt-years); alternative categorization of TCDD Ever-employment in TCP or 2,4,5-T manufacture

’t Mannetje et al., 2005 [173]

Collins et al., 2009 [169]

773 Male PCP manufacture workers with exposure to chlorinated dioxins, including 196 with exposure to TCP, employed in one U.S. plant; serum TCDD level available for 128 workers

Collins et al., 2009 [168]

1615 Male TCP manufacture workers with exposure to TCDD employed in one U.S. plant, including 196 PCP production workers; serum TCDD level available for 280 workers 1599 Male and female TCP production workers in New Zealand

Employment during 1942e1982; followup 1942e2003

NHL, HL, and other lymphopoietic cancer (excluding leukemia and aleukemia) mortality from death certificates

Employment during 1969e1988; follow-up 1969e2004

NHL, HL, and MM mortality from national registry

2106 Male workers in two Dutch plants, one with 2,4,5-T as main product (factory A) and the other with 4-chloro-2methylphenoxyacetic acid, 4-chloro-3methylphenoxy propanoic acid, and 2,4-D as main products (factory B); serum TCDD level available for 187 workers (101 from factory A, 86 from factory B)

Employment 1955e1985 (plant A), 1965e1986 (plant B); follow-up to 2006

NHL mortality from national registry

McBride et al., 2009 [171]

Boers et al., 2010 and 2012 [166, 167]

Exposure to TCDD defined as (1) employment during manufacture of spraying of 2,4,5-T, PCP or other potentially contaminated compounds or (2) employment in plants with documented (biomonitoring or environmental measurements) exposure to TCDD Biomonitoring data available for 573 workers from 10 cohorts 13,831 Workers (63.3%) classified as exposed to TCDD Ever-employment in the plants Biomonitoring data available for 253 workers

(continued on next page)

292.e5

Employment during 2,4,5-T production; 1134 workers classified as TCDD exposed and 346 as unexposed Employment in selected departments (or entering them on a regular basis), or exposure to an accident in factory A in 1963 Predictive model for TCDD plasma levels at time of assumed last exposure derived by two-stage regression model, with exposure status based on detailed occupational history or using a priori assumed exposure status based on department of employment

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Ott and Zober, 1996 [174]

Author, publication year Ruder and Yiin, 2011 [172] Manuwald et al., 2012 [170] Studies of Seveso, Italy Consonni et al., 2008 [182]

Warner et al., 2011 [184]

Community-based studies Hardell et al., 1995 [186, 185]

Period of exposure and follow-up

Definition of outcome

Definition of exposure

2122 Workers at four U.S. PCP production plants 1489 Production workers in Germany

Employment from 1930s until 1980s, one plant until 2005; follow-up 1940e2005 Employment 1952e1984; follow-up 1952e2007

NHL, HL, and MM mortality from death certificates NHL mortality from death certificates obtained from local population registries

Employment in TCP production (exposure to TCDD, n ¼ 720) Ever-employment (minimum 3 mo) in the plant

723 Residents in zone A (“very high contamination”), 4821 residents in zone B (“high contamination”), 31,643 residents in zone R (“low contamination”), and 181,574 residents in reference area (“unaffected”) based on their official residence on the day of the industrial accident

Exposure on the day of accident on July 10, 1976; follow-up 1976e2001

NHL, HL, or MM mortality from death certificates

723 Residents in zone A (“very high contamination”), 4821 in zone B (“high contamination”), 31,643 in zone R (“low contamination”), and 181,574 in reference area (“unaffected”) based on their official residence on the day of the industrial accident 981 Women aged 0e40 y and residing in zone A or B in 1976, with adequate stored sera collected soon after the industrial accident; enrolled in 1996e1998, followed up in 2008e2009 Participation: 1271 eligible in 1996e1997, 33 deceased or ill, 17 not reachable, 240 refused, and 981 enrolled; in 2008e2009: 16 deceased, 36 not reachable, 96 refused, and 833 participated

Exposure on the day of accident on July 10, 1976; follow-up 1976e1996

Incidence of NHL, HL, or MM from cancer registry

Residence in contaminated area in 1976 Range of mean TCDD levels in soil (mg/ m2): Zone A: 15.5e580.4 Zone B: 1.7e4.3 Zone R: 0.9e1.4 Reference: NA Median lipid-adjusted TCDD levels in serum of residents (pg/g or ppt): Zone A: 447.0 in 1976e1977, 73.3 in 1992e1996 Zone B: 94.0 in 1976e1977, 12.4 in 1992e1996 Zone R: 48.0 in 1976e77, NA in 1992e96 Reference: NA in 1976e1977, 5.5 in 1992e1996 Residence in contaminated area in 1976

Exposure at accident on July 10, 1976; follow-up 1976e2009

Incidence of doctor-diagnosed lymphoma self-reported in personal interview, confirmed by review of medical records; death certificates reviewed for 16 deaths, medical records obtained for confirmation if cancer due to cancer

Residence in zone A or B in 1976

Seven male patients (mean age ¼ 60 y, range ¼ 49e75) with histopathologically confirmed NHL (n ¼ 6) or plasmacytoma (n ¼ 1) and one male HL patient attending a single oncology department; 12 elective surgery patients at the same hospital (mean age ¼ 55 y, range ¼ 35e76 in six men; mean age ¼ 49 y, range ¼ 23e67 in six women) with benign gallbladder disease (n ¼ 11) or epigastric hernia (n ¼ 1) and without diagnosed malignant disease or known exposure to polychlorinated dibenzodioxins or dibenzofurans

Tissue samples collected in 1989e1991

Malignant lymphoproliferative disease, histopathologically confirmed, including 11 NHLs, 1 plasmacytoma, and 1 HL

TCDD and TEQ for polychlorinated dibenzodioxins and dibenzofurans measured in w10 g adipose tissue obtained from the abdominal wall; TCDD not measured in two patients, in whom TEQ was calculated using other congeners

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Pesatori et al., 2009 [183]

Study subjects

292.e6

Supplementary Table 1 (continued )

Hardell et al., 1996, 2001 [188, 187]

Kramárová et al., 1998, 2004 [190, 191]

Cases and controls recruited in 1994e1997

Newly diagnosed NHL at a single oncology department; stratified between high-grade and low-grade B-cell NHL according to the Kiel classification

TCDD, other polychlorinated dibenzodioxins, dibenzofurans, and biphenyls measured in 2- to 10-g adipose tissue obtained from the abdominal wall; analytical laboratory blinded to caseecontrol status Immunoglobulin G (IgG) and IgM antibodies measured against EBV viral capsid antigen and combined restricted and diffuse early antigen; IgG measured against EBV nuclear antigen 1

Cases and controls enrolled starting in 1993

NHL histopathologically reviewed by independent pathologists in Vietnam and Europe

Self-reported personal experience of aerial herbicide spraying during the Vietnam War [192] Exposure Opportunity Index calculated based on self-reported residential history and U.S. Army HERBS tape information on patterns and extent of herbicide use in Vietnam TCDD measured in 2-g adipose tissue collected from 25 subjects; 20-mL blood also collected for TCDD analysis, but results were not discussed in 2004 abstract

Cases and controls enrolled in June 1998 through June 2000

Newly diagnosed NHL ascertained from one of four U.S. Surveillance, Epidemiology and End Results cancer registries, with histologic subtype information based on abstracted reports of the diagnosing pathologist, coded according to International Classification of Disease for Oncology, Second Edition, and recoded to the World Health Organization classification

Concentrations of TCDD, 6 other dioxins, 10 furans, 13 organochlorine pesticides or pesticide metabolites, and 36 noncoplanar and 4 coplanar polychlorinated biphenyl congeners measured in plasma; TEQ calculated as the sum of lipid-adjusted levels of dioxin, furan, and relevant polychlorinated biphenyl congeners, each weighted by its congener-specific toxic equivalency factor relative to TCDD; blood samples provided by 62% of cases and 66% of controls

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

De Roos et al., 2005 [193]

33 Patients with newly diagnosed, previously untreated NHL and no prior malignancy, recruited from a single oncology department with 100% participation; comparison group of 39 consecutive surgery patients with benign lesions and no history of malignancy, frequency matched on age, recruited from surgery department at the same hospital with 100% participation EBV antibody titers measured in 23 cases and 32 controls 150 Histopathologically confirmed NHL cases identified from the Ho Chi Minh City Cancer Center and w300 hospital controls matched to cases by age, sex, and place of residence (Ho Chi Minh City or elsewhere), including half with cancer (other than liver cancer, respiratory organ cancer, prostate cancer, thyroid and endocrine tumors, HL, leukemias, and MM) and half with another condition (usually a benign tumor; w50% referred to cancer center for a suspected malignancy that was not pathologically confirmed); effective sample size ¼ 124 matched triplets; participation rates not reported 100 Untreated NHL cases and 100 randomly selected population controls from a larger NHL caseecontrol study based in Iowa, Los Angeles County, Detroit area, and Seattle area, including newly diagnosed NHL cases aged 20e74 years without evidence of HIV infection and controls identified by random-digit dialling (if aged <65 y) or from Medicare eligibility files (if aged 65 y) frequency matched to cases by age, sex, and race, enrolled between July 1998 and June 2000; response rates ¼ 59% for cases and 44% for controls, participation rates among eligible contacted subjects ¼ 76% for cases and 52% for controls

292.e7

Author, publication year Air Force Health Study Ketchum and Akhtar, 1996 [114]; Michalek et al., 1998 [112]

Ketchum and Michalek, 2005 [115]

Army chemical corps study Thomas and Kang, 1990 [120]

Cypel and Kang, 2010 [121]

Level of exposure (with N cases)

RR for NHL (95% CI)

RR for HL (95% CI)

RR for MM (95% CI)

Ranch Hands 0 Reticulosarcoma deaths 0 Reticulolymphosarcoma deaths 1 Other lymphoma death 0 CLL deaths 0 HL deaths 1 MM death Expected based on comparison subjects 0.179 Reticulosarcoma deaths 0.092 Reticulolymphosarcoma deaths 0.452 Other lymphoma deaths 0.268 CLL deaths 0.140 HL deaths 0.865 MM deaths Ranch Hands 2 NHLs (1 with background TCDD, 1 with high TCDD) 5 CLLs (3 with background TCDD, 1 with low TCDD, 1 with unknown TCDD) 1 HL (background TCDD) 4 MMs (1 with background TCDD, 3 with low TCDD) Comparison subjects 7 NHLs (5 with TCDD 10 ppt, 1 with TCDD >10 ppt, 1 with unknown TCDD) 6 CLLs (6 with TCDD 10 ppt) 3 HLs (3 with TCDD 10 ppt) 1 MM (1 with TCDD 10 ppt) Ranch Hands 1 Malignant lymphoma death 0 NHL deaths 0 HL deaths 0 Lymphoreticulosarcoma deaths 2 MM deaths

Reticulosarcoma SMR ¼ 0 (0e20.6) Reticulolymphosarcoma SMR ¼ 0 (0e40.1) Other lymphoma SMR ¼ 2.2 (0.06e12.3) CLL SMR ¼ 0 (0e13.8)

HL SMR ¼ 0 (0e26.4)

MM SMR ¼ 1.2 (0.03e6.4)

NR

NR

NR

NR

NR

NR

NR NHL SMR ¼ 0 (0e9.2)

NR Agent Orange Registry: HL SIR ¼ 5.0 (0.1e27.9) Overall: HL SIR ¼ 2.9 (0.3e10.3)

NR

Lymphoid cancer RR versus non-Vietnam Army Chemical Corps veterans ¼ 1.93 (0.35e10.51) Vietnam Army Chemical Corps: lymphoid cancer SMR ¼ 0.48 (0.13e1.24) Non-Vietnam Army Chemical Corps: lymphoid cancer SMR ¼ 0.33 (0.04e1.20)

NR

NR

Army Chemical Corps 0 NHL cases 1 HL case (confirmed) in Agent Orange Registry 1 HL case (confirmed) in VA Patient Treatment File (0.2 HL cases expected in Agent Orange Registry, 0.7 expected overall) 0 NHL deaths (0.4 expected) 0 HL deaths Vietnam Army Chemical Corps: 4 lymphoid cancer deaths Non-Vietnam Army Chemical Corps: 2 lymphoid cancer deaths

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

AFHS, 2005 [107]

292.e8

Supplementary Table 2 Results of epidemiologic studies of veterans with estimated Agent Orange/TCDD exposure

Selected Cancers Study Selected Cancers Cooperative Study Group, 1990 [123, 122]

Military service in Vietnam (99 NHLs, 28 HLs) Military service in Vietnam in 1964e1972 (96 NHLs, 26 HLs)

Vietnam veterans Army (45 NHLs, 12 HLs) Air Force (12 NHLs, 5 HLs) Marines (10 NHLs, 4 HLs) Navy (32 NHLs, 7 HLs) Coast Guard (0 NHLs, 0 HLs) Non-Vietnam veterans Army (239 NHLs, 38 HLs)

Marines (24 NHLs, 12 HLs) Navy (71 NHLs, 24 HLs) Coast Guard (5 NHLs, 0 HLs) No military service in Vietnam (1058 NHLs, 282 HLs) Duration of service in Vietnam <1 y (42 NHLs, 13 HLs) 1e1.4 y (18 NHLs, 8 HLs) 1.5e1.9 y (18 NHLs, 3 HLs) 2 y (18 NHLs, 4 HLs) Calendar years in Vietnam Before 1966 (9 NHLs, 1 HL) 1966e1969 (74 NHLs, 19 HLs) After 1969 (14 NHLs, 8 HLs) Age at the first tour in Vietnam <21 y (42 NHLs, 15 HLs) 21e25 y (31 NHLs, 9 HLs) 26 y (25 NHLs, 4 HLs) Rank at the end of the last tour in Vietnam E1eE3 (12 NHLs, 5 HLs) E4eE9 (72 NHLs, 21 HLs) Officer (12 NHLs, 2 HLs) Type of unit in Vietnam Support (56 NHLs, 20 HLs)

HL OR ¼ 1.14 (0.71e1.83) versus men who did not serve in Vietnam HL OR ¼ 1.09 (0.62e1.91) versus non-Vietnam veterans HL OR ¼ 1.17 (0.71e1.92) versus non-veterans HL OR ¼ 1.23 (0.65e2.36) versus non-Vietnam Vietnam-era Veterans

NHL OR ¼ 1.19 (0.79e1.80) versus nonveterans NHL OR ¼ 1.02 (0.47e2.24) versus nonveterans NHL OR ¼ 1.84 (0.78e4.34) versus nonveterans NHL OR ¼ 1.89 (1.11e3.24) versus nonveterans NHL OR ¼ NR; Pheterogeneity ¼ .29

HL OR ¼ 1.00 (0.51e1.97)

NHL OR ¼ 0.82 nonveterans NHL OR ¼ 0.91 nonveterans NHL OR ¼ 0.77 nonveterans NHL OR ¼ 0.79 nonveterans NHL OR ¼ 1.45 nonveterans Referent

NR

HL OR ¼ 1.69 (0.58e4.89) HL OR ¼ 1.73 (0.51e5.88) HL OR ¼ 1.06 (0.43e2.59) HL OR ¼ NR

(0.66e1.02) versus

HL OR ¼ 0.78 (0.52e1.17)

(0.67e1.25) versus

HL OR ¼ 1.36 (0.81e2.29)

(0.45e1.32) versus

HL OR ¼ 1.77 (0.86e3.63)

(0.57e1.11) versus

HL OR ¼ 1.34 (0.81e2.24)

(0.39e5.38) versus

HL OR ¼ NR

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Air Force (88 NHLs, 23 HLs)

NHL OR ¼ 1.47 (1.09e1.97) versus men who did not serve in Vietnam NHL OR ¼ 1.63 (1.14e2.33) versus nonVietnam veterans NHL OR ¼ 1.41 (1.03e1.93) versus nonveterans NHL OR ¼ 1.52 (1.00e2.32) versus nonVietnam Vietnam-era veterans

Referent

NHL OR ¼ 1.05 (0.70e1.57) NHL OR ¼ 1.98 (1.00e3.94) NHL OR ¼ 2.99 (1.41e6.31); Pheterogeneity ¼ .06 NHL OR ¼ 1.54 (0.79e3.01); Ptrend ¼ .10

HL OR ¼ 0.86 (0.45e1.65) HL OR ¼ 2.43 (0.98e5.99) HL OR ¼ 1.14 (0.30e4.37)

NHL OR ¼ 1.38 (0.54e3.55) NHL OR ¼ 1.41 (1.01e1.98) NHL OR ¼ 1.64 (0.79e3.39); Pheterogeneity ¼ .93

HL OR ¼ 0.41 (0.05e3.44) HL OR ¼ 1.11 (0.64e1.93) HL OR ¼ 1.63 (0.66e4.02); Pheterogeneity ¼ .42

NHL OR ¼ 1.73 (1.11e2.70) NHL OR ¼ 1.20 (0.74e1.94) NHL OR ¼ 1.45 (0.81e2.60); Pheterogeneity ¼ .52

HL OR ¼ 1.15 (0.61e2.18) HL OR ¼ 1.06 (0.49e2.29) HL OR ¼ 1.31 (0.42e4.10); Pheterogeneity ¼ .95

NHL OR ¼ 1.29 (0.61e2.72) NHL OR ¼ 1.44 (1.02e2.03) NHL OR ¼ 1.78 (0.80e3.96); Pheterogeneity ¼ .84

HL OR ¼ 1.07 (0.37e3.04) HL OR ¼ 1.19 (0.70e2.04) HL OR ¼ 1.33 (0.28e6.26); Pheterogeneity ¼ .97

NHL OR ¼ 1.50 (1.02e2.21)

HL OR ¼ 1.58 (0.90e2.77)

HL OR ¼ 1.11 (0.36e3.45); Pheterogeneity ¼ .33

(continued on next page)

292.e9

Author, publication year

Level of exposure (with N cases) Combat support (18 NHLs, 3 HLs) Combat (15 NHLs, 5 HLs) Corps in Vietnam I (23 NHLs, 7 HLs) II (19 NHLs, 3 HLs) III (19 NHLs, 9 HLs) IV (2 NHLs, 1 HL) Blue-water Navy (28 NHLs, 7 HLs) Ever in III Corps in Vietnam No (38 NHLs, 10 HLs) Yes (27 NHLs, 11 HLs) Blue-water Navy (28 NHLs, 7 HLs)

Time since the start of the first Vietnam tour <17 y (13 NHLs) 17e18 y (30 NHLs) 19e21 y (39 NHLs) 22 y (16 NHLs) Among Vietnam veterans Passed through defoliated area (27 NHLs, 8 HLs) Any possible contact with Agent Orange (26 NHLs, 7 HLs) Present when others sprayed Agent Orange (9 NHLs, 2 HLs) Agent Orange on skin or clothes (8 NHLs, 1 HL) Handling equipment or containers used with Agent Orange (1 NHL, 0 HLs) Spraying Agent Orange (0 NHLs, 0 HLs) Overall Occupational exposure to phenoxy herbicides (59 NHLs, 16 HLs) Occupational exposure to chlorophenols (140 NHLs, 39 HLs) U.S Department of Veterans Affairs hospitals study Dalager et al., 1991 [124] Service in Vietnam (100 NHLs) Army in Vietnam (66 NHLs) Army, combat duty in Vietnam (23 NHLs) Army, noncombat duty in Vietnam (40 NHLs) Marines in Vietnam (9 NHLs) Marines, combat duty in Vietnam (7 NHLs)

RR for HL (95% CI) HL OR ¼ 0.50 (0.14e1.76) HL OR ¼ 0.94 (0.34e2.59); Pheterogeneity ¼ .17

NHL OR ¼ 2.25 (1.21e4.18) NHL OR ¼ 1.22 (0.66e2.26) NHL OR ¼ 0.89 (0.50e1.58) NHL OR ¼ 0.90 (0.15e5.41); Pheterogeneity ¼ .11 NHL OR ¼ 2.17 (1.22e3.86)

HL OR ¼ 1.67 (0.67e4.18) HL OR ¼ 0.52 (0.15e1.81) HL OR ¼ 1.25 (0.57e2.75) HL OR ¼ 0.93 (0.09e9.82); Pheterogeneity ¼ .59 HL OR ¼ 1.39 (0.56e3.46)

NHL OR ¼ 1.70 (1.07e2.71) NHL OR ¼ 0.96 (0.59e1.57) NHL OR ¼ 2.18 (1.23e3.88); Pheterogeneity ¼ .06

HL OR ¼ 1.27 (0.60e2.67) HL OR ¼ 1.12 (0.55e2.27) HL OR ¼ 1.42 (0.57e3.52); Pheterogeneity ¼ .92

NHL OR ¼ 1.30 (0.93e1.82) NHL OR ¼ 1.29 (0.92e1.82) NHL OR ¼ 2.26 (0.52e9.78) NHL OR ¼ NR NHL OR ¼ 2.18 (1.23e3.87); Pheterogeneity ¼ .11 for all land-based versus sea-based

HL OR ¼ 1.08 (0.64e1.82) HL OR ¼ 1.18 (0.70e2.00) HL OR ¼ NR HL OR ¼ NR HL OR ¼ 1.41 (0.57e3.50); Pheterogeneity ¼ .61 for all land-based versus sea-based

NHL NHL NHL NHL

OR OR OR OR

¼ ¼ ¼ ¼

1.35 1.51 1.40 1.63

(0.64e2.82) (0.91e2.51) (0.90e2.18) (0.80e3.32); Ptrend ¼ .75

RR for MM (95% CI)

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Land versus sea duty in Vietnam All land based (71 NHLs, 21 HLs) Land-based non-Navy (67 NHLs, 21 HLs) Navy, shore (4 NHLs, 0 HLs) Navy, brown-water (0 NHLs, 0 HLs) Sea-based blue-water Navy (28 NHLs, 7 HLs)

RR for NHL (95% CI) NHL OR ¼ 1.18 (0.65e2.15) NHL OR ¼ 1.25 (0.63e2.45); Pheterogeneity ¼ .76

HL OR ¼ NR

NHL OR ¼ 0.82 (0.45e1.49)

HL OR ¼ 0.65 (0.25e1.67)

NHL OR ¼ 1.08 (0.58e2.02)

HL OR ¼ 0.76 (0.28e2.07)

NHL OR ¼ 0.98 (0.39e2.48)

HL OR ¼ 0.74 (0.15e3.67)

NHL OR ¼ 1.08 (0.40e2.96)

HL OR ¼ 0.47 (0.06e3.99)

NHL OR ¼ 0.41 (0.04e4.07)

HL OR ¼ NR

NHL OR ¼ NR

HL OR ¼ NR

P  .05 (5.1% cases, 5.9% controls)

P  .05 (5.2% cases, 5.9% controls)

P  .05 (12.1% cases, 11.3% controls)

P  .05 (12.6% cases, 11.3% controls)

NHL OR versus never in Vietnam ¼ 1.03 (0.70e1.50), adjusted for military branch NHL OR versus Army non-Vietnam ¼ 1.13 (0.72e1.78) NHL OR versus Army non-Vietnam ¼ 0.91 (0.50e1.65) NHL OR versus Army non-Vietnam ¼ 1.27 (0.76e2.14) NHL OR versus Marines non-Vietnam ¼ 0.83 (0.25e2.72) NHL OR versus Marines non-Vietnam ¼ 0.89 (0.25e3.08)

NR

292.e10

Supplementary Table 2 (continued )

NR

Marines, noncombat duty in Vietnam (2 NHLs) Navy in Vietnam (19 NHLs) Air Force in Vietnam (6 NHLs) Army, I Corps (14 NHLs) Army, II Corps (31 NHLs) Army, III Corps (26 NHLs) Army, IV Corps (3 NHLs) Army, 5e10 y since Vietnam entry (15 NHLs) Army, 11e15 y since Vietnam entry (24 NHLs) Army, >15 y since Vietnam entry (25 NHLs)

Navy, 11e15 y since Vietnam entry (7 NHLs) Navy, >15 y since Vietnam entry (7 NHLs)

Dalager et al., 1995 [125]

NR

Referent HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼ HL OR ¼

NR 1.28 1.08 1.15 1.73 1.09 1.27 1.31 1.00 1.66 0.77 1.28 1.34 1.22 1.51 1.65 1.20 0.51 1.18 1.07 1.25 0.64

(0.94e1.76) (0.90e1.30) (0.94e1.42) (0.68e4.44) (0.93e1.28) (0.67e2.43) (0.91e1.91) (0.52e0.92) (0.63e4.38) (0.23e2.59) (0.81e2.03) (0.90e2.01) (0.87e1.70) (0.88e2.59) (1.02e2.68) (0.85e1.71) (0.10e2.64) (0.68e2.05) (0.65e1.75) (0.78e2.00) (0.22e1.92) (continued on next page)

292.e11

Air Force, 5e10 y since Vietnam entry (1 NHL) Air Force, 11e15 y since Vietnam entry (2 NHLs) Air Force, >15 y since Vietnam entry (2 NHLs) Never in Vietnam (141 HLs) Service in Vietnam (142 HLs) Army in Vietnam (86 HLs) Marines in Vietnam (23 HLs) Air Force in Vietnam (10 HLs) Navy in Vietnam (23 HLs) <6 m in Vietnam (20 HLs) 6e11 m in Vietnam (82 HLs) 12e17 m in Vietnam (17 HLs) 18e23 m in Vietnam (9 HLs) 24 m in Vietnam (4 HLs) Army/Marines combat in Vietnam (43 HLs) Army/Marines support in Vietnam (66 HLs) Ground duty in Vietnam (111 HLs) Other duty in Vietnam (33 HLs) Grades E1eE3 (44 HLs) Grades E4eE9 (96 HLs) Officer (2 HLs) Army, I Corps (27 HLs) Army, II Corps (34 HLs) Army, III Corps (40 HLs) Army, IV Corps (5 HLs)

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Marine Corps, 5e10 y since Vietnam entry (4 NHLs) Marine Corps, 11e15 y since Vietnam entry (2 NHLs) Marine Corps, >15 y since Vietnam entry (3 NHLs) Navy, 5e10 y since Vietnam entry (4 NHLs)

NHL OR versus Marines non-Vietnam ¼ 0.80 (0.15e4.28) NHL OR versus Navy non-Vietnam ¼ 0.70 (0.31e1.60) NHL OR versus Air Force non-Vietnam ¼ 1.40 (0.42e4.72) NHL OR versus Army non-Vietnam ¼ 1.20 (0.58e2.50) NHL OR versus Army non-Vietnam ¼ 1.62 (0.91e2.89) NHL OR versus Army non-Vietnam ¼ 0.70 (0.40e1.23) NHL OR versus Army non-Vietnam ¼ 0.41 (0.12e1.34) NHL OR versus Army non-Vietnam ¼ 1.28 (0.62e2.64) NHL OR versus Army non-Vietnam ¼ 1.00 (0.56e1.82) NHL OR versus Army non-Vietnam ¼ 1.14 (0.63e2.06) NHL OR versus Marines non-Vietnam ¼ 1.44 (0.34e6.19) NHL OR versus Marines non-Vietnam ¼ 0.58 (0.11e3.01) NHL OR versus Marines non-Vietnam ¼ 0.81 (0.18e3.66) NHL OR versus Navy non-Vietnam ¼ 0.72 (0.18e2.84) NHL OR versus Navy non-Vietnam ¼ 1.47 (0.40e5.39) NHL OR versus Navy non-Vietnam ¼ 0.70 (0.23e2.12) NHL OR versus Air Force non-Vietnam ¼ 1.40 (0.14e14.35) NHL OR versus Air Force non-Vietnam ¼ 0.64 (0.13e3.12) NHL OR versus Air Force non-Vietnam ¼ NR

Author, publication year

Level of exposure (with N cases)

RR for NHL (95% CI)

Vietnam before 1967 (20 HLs) Vietnam 1967e1969 (88 HLs) Vietnam after 1969 (34 HLs) <6 y since Vietnam entry (10 HLs) 6e10 y since Vietnam entry (52 HLs) 11e15 y since Vietnam entry (54 HLs) >15 y since Vietnam entry (19 HLs) Korean veterans study Yi et al., 2013 [126]

Self-reported perceived exposure Low: 22 NHLs, 22 HLs, 213 MMs High: 59 NHLs, 45 HLs, 607 MMs

Battalion/company-level exposure Low: 46 NHLs, 41 HLs, 377 MMs High: 28 NHLs, 19 HLs, 322 MMs

Yi et al., 2014 [127]

Modeled exposure None: 33 NHLs, 3 HLs, 8 MMs Low: 56 NHLs, 3 HLs, 15 MMs Moderate: 58 NHLs, 6 HLs, 18 MMs High: 38 NHLs, 1 HL, 10 MM

Self-perceived exposure High versus low: NHL OR ¼ 2.33 (1.43e3.82) Low versus none: NHL OR ¼ 1.47 (0.62e3.52) Moderate versus none: NHL OR ¼ 2.52 (1.37e4.66) High versus none: NHL OR ¼ 2.93 (1.47e5.82) P for trend <.001 Division/brigade-level modeled exposure High versus low: NHL OR ¼ 0.83 (0.51e1.33) Low versus none: NHL OR ¼ 1.49 (0.74e2.97) Moderate versus none: NHL OR ¼ 1.13 (0.55e2.32) High versus none: NHL OR ¼ 0.97 (0.44e2.13) P for trend ¼ .66 Battalion/company-level modeled exposure High versus low: NHL OR ¼ 0.83 (0.52e1.34) Low versus none: NHL OR ¼ 0.98 (0.55e1.76) Moderate versus none: NHL OR ¼ 0.71 (0.35e1.44) High versus none: NHL OR ¼ 0.95 (0.49e1.87) P for trend ¼ .64 High versus low: NHL HR ¼ 1.09 (0.81e1.47) Low versus none: NHL HR ¼ 1.60 (1.04e2.46) Moderate versus none: NHL HR ¼ 1.43 (0.92e2.23) High versus none: NHL HR ¼ 1.42 (0.89e2.28) Per unit increase in log10 exposure: NHL HR ¼ 1.05 (0.98e1.13) Veterans in units with a defined tactical area of responsibility High versus low: NHL HR ¼ 0.99 (0.71e1.39) Low versus none: NHL HR ¼ 1.44 (0.86e2.41) Moderate versus none: NHL HR ¼ 1.26 (0.74e2.14)

OR OR OR OR OR OR OR

¼ ¼ ¼ ¼ ¼ ¼ ¼

1.48 1.14 1.64 2.88 1.24 1.25 1.03

RR for MM (95% CI)

(0.75e2.94) (0.79e1.64) (0.96e2.81) (0.99e8.38) (0.81e1.90) (0.81e1.92) (0.53e2.00)

Self-perceived exposure High versus low: HL OR ¼ 1.84 (1.10e3.08) Low versus none: HL OR ¼ 2.66 (1.15e6.16) Moderate versus none: HL OR ¼ 2.09 (1.02e4.29) High versus none: HL OR ¼ 4.01 (1.92e8.36) P for trend <.001 Division/brigade-level modeled exposure High versus low: HL OR ¼ 1.03 (0.59e1.79) Low versus none: HL OR ¼ 0.99 (0.48e2.06) Moderate versus none: HL OR ¼ 1.31 (0.62e2.76) High versus none: HL OR ¼ 0.68 (0.27e1.68) P for trend ¼ .63 Battalion/company-level modeled exposure High versus low: HL OR ¼ 0.69 (0.40e1.20) Low versus none: HL OR ¼ 1.39 (0.73e2.64) Moderate versus none: HL OR ¼ 0.92 (0.41e2.04) High versus none: HL OR ¼ 0.76 (0.32e1.81) P for trend ¼ .38 High versus low: HL HR ¼ 1.26 (0.41e3.93) Low versus none: HL HR ¼ 1.02 (0.21e5.07) Moderate versus none: HL HR ¼ 1.90 (0.45e8.01) High versus none: HL HR ¼ 0.45 (0.05e4.44) Per unit increase in log10 exposure: HL HR ¼ 1.07 (0.82e1.40) Veterans in units with a defined tactical area of responsibility High versus low: HL HR ¼ 0.84 (0.22e3.12) Low versus none: HL HR ¼ 0.95 (0.16e5.86) Moderate versus none: HL HR ¼ 1.30 (0.22e7.73)

Self-perceived exposure High versus low: MM OR ¼ 2.45 (2.10e2.87) Low versus none: MM OR ¼ 1.17 (0.87e1.57) Moderate versus none: MM OR ¼ 2.05 (1.68e2.49) High versus none: MM OR ¼ 3.69 (3.02e4.52) P for trend <.001 Division/brigade-level modeled exposure High versus low: MM OR ¼ 1.10 (0.93e1.29) Low versus none: MM OR ¼ 0.91 (0.72e1.16) Moderate versus none: MM OR ¼ 0.97 (0.77e1.22) High versus none: MM OR ¼ 1.14 (0.90e1.45) P for trend ¼ .18 Battalion/company-level modeled exposure High versus low: MM OR ¼ 1.14 (0.98e1.33) Low versus none: MM OR ¼ 1.03 (0.84e1.27) Moderate versus none: MM OR ¼ 1.09 (0.87e1.37) High versus none: MM OR ¼ 1.25 (0.99e1.56) P for trend ¼ .048 High versus low: MM HR ¼ 1.14 (0.65 e2.01) Low versus none: MM HR ¼ 1.80 (0.76 e4.27) Moderate versus none: MM HR ¼ 1.69 (0.72 e3.96) High versus none: MM HR ¼ 1.48 (0.58 e3.78) Per unit increase in log10 exposure: MM HR ¼ 1.09 (0.95e1.26) Veterans in units with a defined tactical area of responsibility High versus low: MM HR ¼ 0.89 (0.47 e1.67) Low versus none: MM HR ¼ 1.24 (0.47e3.29) Moderate versus none: MM HR ¼ 0.96 (0.36e2.58)

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Division/brigade-level modeled exposure Low: 37 NHLs, 30 HLs, 291 MMs High: 37 NHLs, 30 HLs, 408 MMs

RR for HL (95% CI) HL HL HL HL HL HL HL

292.e12

Supplementary Table 2 (continued )

HR ¼ hazard ratio; NR ¼ not reported.

High versus none: HL HR ¼ 0.00 (NR) Per unit increase in log10 exposure: HL HR ¼ 0.98 (0.70e1.37) Veterans who served in Vietnam for >6 months High versus low: HL HR ¼ 1.03 (0.34e3.12) Low versus none: HL HR ¼ 0.90 (0.18e4.45) Moderate versus none: HL HR ¼ 1.39 (0.34e5.63) High versus none: HL HR ¼ 0.36 (0.04e3.45) Per unit increase in log10 exposure: HL HR ¼ 1.01 (0.78e1.32) Veterans of enlisted rank High versus low: HL HR ¼ 0.93 (0.15e5.85) Low versus none: HL HR ¼ 1.72 (0.15e19.4) Moderate versus none: HL HR ¼ 2.46 (0.18e33.8) High versus none: HL HR ¼ 0.00 (NR) Per unit increase in log10 exposure: HL HR ¼ 1.09 (0.68e1.73)

High versus none: MM HR ¼ 1.16 (0.40e3.33) Per unit increase in log10 exposure: MM HR ¼ 1.03 (0.87e1.22) Veterans who served in Vietnam for >6 months High versus low: MM HR ¼ 1.18 (0.66e2.09) Low versus none: MM HR ¼ 2.26 (0.87e5.83) Moderate versus none: MM HR ¼ 2.05 (0.81e5.19) High versus none: MM HR ¼ 1.78 (0.65e4.89) Per unit increase in log10 exposure: MM HR ¼ 1.11 (0.96e1.29) Veterans of enlisted rank High versus low: MM HR ¼ 1.72 (0.77e3.83) Low versus none: MM HR ¼ 2.52 (0.68e9.33) Moderate versus none: MM HR ¼ 3.39 (0.87e13.2) High versus none: MM HR ¼ 3.03 (0.76e12.0) Per unit increase in log10 exposure: MM HR ¼ 1.21 (0.98e1.49)

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

High versus none: NHL HR ¼ 1.30 (0.74e2.30) Per unit increase in log10 exposure: NHL HR ¼ 1.03 (0.95e1.13) Veterans who served in Vietnam for >6 months High versus low: NHL HR ¼ 1.02 (0.76 e1.38) Low versus none: NHL HR ¼ 1.47 (0.95e2.28) Moderate versus none: NHL HR ¼ 1.25 (0.81e1.95) High versus none: NHL HR ¼ 1.30 (0.81e2.08) Per unit increase in log10 exposure: NHL HR ¼ 1.03 (0.96e1.11) Veterans of enlisted rank High versus low: NHL HR ¼ 1.00 (0.66e1.49) Low versus none: NHL HR ¼ 1.87 (1.11e3.15) Moderate versus none: NHL HR ¼ 1.56 (0.85e2.87) High versus none: NHL HR ¼ 1.46 (0.78e2.73) Per unit increase in log10 exposure: NHL HR ¼ 1.06 (0.96e1.17)

292.e13

Author, publication year

Level of exposure (with N cases)

RR for NHL (95% CI)

RR for HL (95% CI)

RR for MM (95% CI)

Kogevinas et al., 1995 [180]

Ever exposed to TCDD (11 NHLs) No TCDD exposure (21 NHLs) Low TCDD exposure (4 NHLs) Medium TCDD exposure (3 NHLs) High TCDD exposure (4 NHLs)

NR

NR

Ott and Zober, 1996 [174]

Exposed: 0 NHL deaths, 0 NHL cases

NR

NR

Kogevinas et al., 1997 [164]

Exposed: 24 NHL deaths, 8 HL deaths, 9 MM deaths Years since the first exposure 0e9: 2 NHL deaths 10e19: 8 NHL deaths 20: 14 NHL deaths Duration of exposure (y) <1: 11 NHL deaths 1e4: 6 NHL deaths 5e9: 3 NHL deaths 10e19: 2 NHL deaths 20: 2 NHL deaths Year of the first exposure <1955: 12 NHL deaths 1955e1964: 9 NHL deaths 1965e1974: 3 NHL deaths 1975: 0 NHL deaths Job title Main production: 9 NHL deaths Maintenance: 1 NHL death Other exposed job: 10 NHL deaths Unspecified task: 0 NHL deaths Sprayer: 4 NHL deaths Unexposed: 9 NHL deaths, 1 HL death, 8 MM deaths Years since the first exposure 0e9: 0 NHL deaths 10e19: 4 NHL deaths 20: 5 NHL deaths Duration of exposure (y) <1: 6 NHL deaths 1e4: 0 NHL deaths 5e9: 1 NHL death 10e19: 2 NHL deaths 20: 0 NHL deaths Year of the first exposure: <1955: 4 NHL deaths 1955e1964: 3 NHL deaths 1965e1974: 2 NHL deaths 1975: 0 NHL deaths Job title Main production: 1 NHL death Maintenance: 4 NHL deaths Other exposed job: 3 NHL deaths Unspecified task: 0 NHL deaths Sprayer: 1 NHL death Ever employed: 12 NHL deaths, 3 HL deaths, 10 MM deaths

NHL OR ¼ 1.93 (0.74e5.07) NHL OR ¼ referent NHL OR ¼ 1.42 (0.44e4.64) NHL OR ¼ 3.63 (0.71e18.67) NHL OR ¼ 3.56 (0.66e19.20) Ptrend ¼ .1 SMR ¼ 0 SIR ¼ 0 NHL SMR ¼ 1.39 (0.89e2.06)

HL SMR ¼ 1.29 (0.56e2.53) MM SMR ¼ 1.21 (0.55e2.29)

NHL SMR ¼ 0.63 (0.08e2.26) NHL SMR ¼ 1.45 (0.63e2.87) NHL SMR ¼ 1.63 (0.89e2.73) NHL NHL NHL NHL NHL

SMR SMR SMR SMR SMR

¼ ¼ ¼ ¼ ¼

2.43 1.20 1.14 0.66 1.21

(1.21e4.35) (0.44e2.62) (0.23e3.32) (0.08e2.37) (0.15e4.38)

NHL NHL NHL NHL

SMR SMR SMR SMR

¼ ¼ ¼ ¼

1.49 1.65 0.96 0.00

(0.77e2.60) (0.75e3.12) (0.20e2.80) (0.00e5.95)

NHL NHL NHL NHL NHL NHL

SMR SMR SMR SMR SMR SMR

¼ ¼ ¼ ¼ ¼ ¼

1.33 0.46 4.85 0.00 0.73 1.00

(0.61e2.52) (0.01e2.56) (2.33e8.93) (0.00e4.34) (0.20e1.88) (0.46e1.90)

HL SMR ¼ 0.27 (0.01e1.51) MM SMR ¼ 1.59 (0.69e3.13)

NHL SMR ¼ 0.00 (0.00e2.53) NHL SMR ¼ 1.79 (0.49e4.57) NHL SMR ¼ 0.94 (0.31e2.20) NHL NHL NHL NHL NHL

SMR SMR SMR SMR SMR

¼ ¼ ¼ ¼ ¼

1.45 0.00 0.98 2.74 0.00

(0.53e3.15) (0.00e1.40) (0.02e5.46) (0.33e9.90) (0.00e9.00)

NHL NHL NHL NHL

SMR SMR SMR SMR

¼ ¼ ¼ ¼

0.89 1.40 1.10 0.00

(0.24e2.28) (0.29e4.08) (0.13e3.99) (0.00e6.96)

NHL NHL NHL NHL NHL NHL

SMR SMR SMR SMR SMR SMR

¼ ¼ ¼ ¼ ¼ ¼

0.46 1.81 1.90 0.00 0.35 1.10

(0.01e2.58) (0.49e4.63) (0.39e5.55) (0.00e19.42) (0.01e1.94) (0.56e1.91)

HL SMR ¼ 1.09 (0.22e3.19) MM SMR ¼ 2.07 (0.99e3.80)

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Steenland et al., 1999 [165]

292.e14

Supplementary Table 3 Results of epidemiologic studies of manufacturers and sprayers of herbicides potentially contaminated with TCDD

*

0.4 ppt TCDD: 2/1 NHL deaths

0.4e4.1 ppt TCDD: 1/3 NHL deaths* 4.1e20.1 ppt TCDD: 2/2 NHL deaths* 20.1 ppt TCDD: 2/1 NHL deaths*

NHL SMR ¼ 0.87 (0.02e4.87)

NHL SMR ¼ 2.4 (1.0e4.7)

HL SMR ¼ 5.58 (0.14e31.0) MM/other immunoproliferative neoplasm SMR ¼ 5.51 (1.14e16.1) HL SMR ¼ 0.00 (0.00e16.1) MM/other immunoproliferative neoplasm SMR ¼ 0.00 (0.00e5.29) HL SMR ¼ 0.0 (0.0e6.4) Other lymphopoietic cancer SMR ¼ 1.3 (0.2e4.6)

NHL SMR ¼ 2.8 (1.1e5.7)

HL SMR ¼ 0.0 (0.0e8.2)

Other lymphopoietic cancer SMR ¼ 1.7 (0.2e6.0)

HL SMR ¼ 1.8 (0.2e6.4)

Other lymphopoietic cancer SMR ¼ 0.6 (0.1e2.3)

NHL SMR ¼ 1.3 (0.6e2.6)

HL SMR ¼ 2.0 (0.2e2.3)

Other lymphopoietic cancer SMR ¼ 0.7 (0.1e2.6)

NHL HR ¼ 1.011 (0.961e1.064) NHL SMR ¼ 1.6 (0.3e4.7)

HL SMR ¼ 4.2 (0.1e23.3)

MM SMR ¼ 2.2 (0.2e8.1)

NHL SMR ¼ 1.6 (0.0e8.7)

HL SMR ¼ 0.0 (0.0e47.1)

MM SMR ¼ 0.0 (0.0e12.2)

NR

NR

NHL SMR ¼ 0.69 (0.02e3.84)

NHL NHL NHL NHL NHL NHL NHL NHL

NHL NHL NHL NHL

SMR ¼ 2.4 (0.5e7.1) SMR ¼ 0.8 (0.0e4.7) SMR ¼ 4.5 (1.2e11.5); Ptrend ¼ .61 RR ¼ 1.006 (0.960e1.054) SMR ¼ 1.6 (0.2e5.7) SMR ¼ 2.8 (0.6e8.1) SMR ¼ 3.1 (0.6e9.1) SMR ¼ 1.3 (0.6e2.5)

SMR SMR SMR SMR

¼ ¼ ¼ ¼

2.3 0.0 2.4 2.1

(0.1e12.8) (0.0e7.1) (0.1e13.2) (0.1e11.5)

NHL HR ¼ 0.92 (0.19e4.47) NHL HR ¼ 0.84 (0.05e13.54) NHL HR ¼ 2.16 (0.29e16.18) NHL HR ¼ 0.41 (0.05e3.25)

NHL HR ¼ NR NHL HR ¼ 1.36 (1.06e1.74) Factory A NHL HR ¼ 1.27 (0.95e1.71) Empirical model referent A priori model referent Empirical NHL HR ¼ 2.99 (0.21e43.29) A priori NHL HR ¼ 3.77 (0.42e34.27) Empirical NHL HR ¼ 5.28 (0.48e58.06) A priori NHL HR ¼ 7.78 (0.68e89.27) Empirical NHL HR ¼ 10.28 (1.05e100.4) A priori NHL HR ¼ 8.09 (0.44e149.07)

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

’t Mannetje et al., 2005 [173] Production workers: 1 NHL death, 1 HL death, 3 MM and other immunoproliferative neoplasm deaths Sprayers: 1 NHL death, 0 HL deaths, 0 MM and other immunoproliferative neoplasm deaths Collins et al., 2009 [169] Whole cohort: 8 NHL deaths, 0 HL deaths, 2 other lymphopoietic cancer deaths Excluding TCP workers: 7 NHL deaths, 2 other lymphopoietic cancer deaths <0.70 ppb-years TEQ: 3 NHL deaths 0.70e3.99 ppb-years TEQ: 1 NHL death 4.00 ppb-years TEQ: 4 NHL deaths Per 1 ppb-year TEQ 0.0e0.14 ppb-year TCDD: 2 NHL deaths 0.15e0.824 ppb-year TCDD: 3 NHL deaths 0.825 ppb-yrs TCDD: 3 NHL deaths Collins et al., 2009 [168] Whole cohort: 9 NHL deaths, 2 HL deaths, 2 other lymphopoietic cancer deaths Excluding PCP workers: 8 NHL deaths, 2 HL deaths, 2 other lymphopoietic cancer deaths Per 1 ppb-year TCDD McBride et al., 2009 [171] Exposed to TCDD: 3 NHL deaths, 1 HL death, 2 MM deaths Unexposed to TCDD: 1 NHL death, 0 HL deaths, 0 MM deaths 0e68.3 ppt-months TCDD: 1 NHL death 68.4e475.0 ppt-months TCDD: 0 NHL deaths 475.1e2085.7 ppt-months TCDD: 1 NHL death >2085.8 ppt-months TCDD: 1 NHL death Boers et al., 2010; Factory A 2012 [166, 167] Exposed: 4 NHL deaths Unexposed: 3 NHL deaths Accident 1963: 1 exposed, 3 unexposed NHL deaths Main production: 2 exposed, 3 unexposed NHL deaths Occasionally exposed: 1 exposed, 3 unexposed NHL deaths Factory B Exposed: 1 NHL death Unexposed: 0 NHL deaths Per 1 unit plasma ln-TCDD, 10-yr lag

(continued on next page)

292.e15

Author, publication year

Level of exposure (with N cases)

RR for NHL (95% CI)

RR for HL (95% CI)

Ruder and Yiin, 2011 [172]

Whole cohort: 17 NHL deaths, 1 HL death, 7 MM deaths Exposed: 8 NHL deaths, 0 HL deaths, 1 MM death Unexposed: 9 NHL deaths, 1 HL death, 6 MM deaths White males: 17 NHL deaths, 0 HL deaths, 6 MM deaths Other males: 0 NHL deaths, 1 HL death, 1 MM death White females: 0 NHL, HL, and MM deaths 57 d of work: 6 NHL deaths (HL and MM NR) 58 to <182 d of work: 4 NHL deaths (HL and MM NR) 182 to <650 d of work: 4 NHL deaths (HL and MM NR) 650 d of work: 3 NHL deaths (HL and MM NR)

NHL SMR ¼ 1.77 (1.03e2.84)

HL SMR ¼ 0.64 (0.02e3.56) MM SMR ¼ 1.50 (0.60e3.10)

NHL SMR ¼ 2.50 (1.08e4.93) NHL SMR ¼ 1.41 (0.64e2.67) NHL SMR ¼ 1.98 (1.15e3.17)

HL SMR ¼ 0.00 (0.00e6.86) MM SMR ¼ 0.72 (0.02e3.99) HL SMR ¼ 0.97 (0.02e5.41) MM SMR ¼ 1.84 (0.68e4.00) HL SMR ¼ 0.00 (0.00e2.63) MM SMR ¼ 1.61 (0.59e3.50)

NHL SMR ¼ 0.00 (0.00e5.33) HL SMR ¼ 7.84 (0.20e43.7) MM SMR ¼ 1.27 (0.03e7.10) NHL SMR ¼ 0.00 (0.00e12.2) HL SMR ¼ 0.00 (0.00e103) MM SMR ¼ 0.00 (0.00e27.5) NHL SMR ¼ 2.45 (0.90e5.34), SRR ¼ referent NHL SMR ¼ 1.56 (0.42e3.99), SRR ¼ 0.55 (0.15e1.97) NHL SMR ¼ 1.63 (0.45e4.18), SRR ¼ 0.63 (0.18e2.28) NHL SMR ¼ 1.42 (0.29e4.14), SRR ¼ 0.62 (0.15e2.55), trend b ¼ 3.744e8 NHL SMR ¼ 1.81 (0.83e3.43) NHL SMR ¼ 2.18 (0.94e4.30) NHL SMR ¼ 0.00 (0.00e7.48) NHL SMR ¼ 0.00 (0.00e7.97) NHL SMR ¼ 1.59 (0.64e3.28) NR NR NHL SMR ¼ 1.56 (0.50e3.65) NHL SMR ¼ 1.67 (0.19e6.02)

HR ¼ hazard ratio; NR ¼ not reported; PCP ¼ pentachlorophenol; TCP ¼ trichlorophenol. * The first N is based on occupational history in different departments; the second N is based on a priori assumed exposure status

Supplementary Table 4 Results of epidemiologic studies of the Seveso industrial accident Author, publication year

Level of exposure (with N cases)

RR for NHL (95% CI)

RR for HL (95% CI)

RR for MM (95% CI)

Consonni et al., 2008 [182]

Zone A: 3 NHL deaths (2 in females), 0 HL deaths, 2 MM deaths Zone B: 7 NHL deaths, 3 HL deaths, 5 MM deaths Zone R: 4 NHL deaths, 9 HL deaths, 24 MM deaths Zone A: 1 NHL, 0 HLs, 1 MM 0e4 years: 0 NHLs, HL NR, 0 MMs 5e9 years: 0 NHLs, HL NR, 0 MMs 10e14 years: 0 NHLs, HL NR, 0 MMs 15 years: 1 NHL, HL NR, 1 MM Zone B: 12 NHLs, 3 HLs, 6 MMs 0e4 years: 2 NHLs, HL NR, 2 MMs 5e9 years: 2 NHLs, HL NR, 2 MMs 10e14 years: 5 NHLs, HL NR, 2 MMs 15 years: 3 NHLs, HL NR, 0 MMs Zone R: 49 NHLs, 23 HLs, 18 MMs 0e4 years: 5 NHLs, HL NR, 1 MM 5e9 years: 15 NHLs, HL NR, 2 MMs 10e14 years: 8 NHLs, HL NR, 4 MMs 15 years: 21 NHLs, HL NR, 11 MMs Lipid-adjusted serum TCDD; 2 NHLs

Zone A NHL RR ¼ 3.35 (1.07e10.46) Females NHL RR ¼ 4.45 (1.10e17.99) Zone B NHL RR ¼ 1.23 (0.58e2.60) Zone R NHL RR ¼ 0.99 (0.71e1.38) Zone A NHL RR ¼ 0.80 (0.11e5.69) NHL RR ¼ NR NHL RR ¼ NR NHL RR ¼ NR NHL RR ¼ 1.97 (0.27e14.07) Zone B NHL RR ¼ 1.51 (0.85e2.69) NHL RR ¼ 1.75 (0.43e7.20) NHL RR ¼ 1.37 (0.34e5.62) NHL RR ¼ 2.30 (0.93e5.66) NHL RR ¼ 0.94 (0.30e2.96) Zone R NHL RR ¼ 0.90 (0.66e1.22) NHL RR ¼ 0.61 (0.24e1.53) NHL RR ¼ 1.44 (0.82e2.53) NHL RR ¼ 0.54 (0.26e1.12) NHL RR ¼ 0.99 (0.62e1.58) NR

Zone A HL RR ¼ 0

Zone A MM RR ¼ 4.34 (1.07e17.52)

Zone B HL RR ¼ 2.15 (0.67e6.86) Zone R HL RR ¼ 0.94 (0.46e1.89) Zone A HL RR ¼ 0

Zone B MM RR ¼ 1.68 (0.69e4.10) Zone R MM RR ¼ 1.10 (0.71e1.69) Zone A MM RR ¼ 2.88 (0.40e20.70) MM RR ¼ NR MM RR ¼ NR MM RR ¼ NR MM RR ¼ 8.35 (1.14e61.31) Zone B MM RR ¼ 2.77 (1.2e6.32) MM RR ¼ 3.56 (0.85e15.00) MM RR ¼ 4.76 (1.11e20.38) MM RR ¼ 4.47 (1.04e19.20) MM RR ¼ NR Zone R MM RR ¼ 1.15 (0.70e1.91) MM RR ¼ 0.23 (0.03e1.70) MM RR ¼ 0.63 (0.15e2.68) MM RR ¼ 1.28 (0.44e3.77) MM RR ¼ 2.24 (1.11e4.49) NR

Pesatori et al., 2009 [183]

Warner et al., 2011 [184] NR ¼ not reported.

Zone B HL RR ¼ 1.20 (0.38e3.78)

Zone R HL RR ¼ 1.46 (0.91e2.29)

NR

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Manuwald et al., 2012 [170]

Sauget, IL: 9 NHL deaths Midland, MI: 8 NHL deaths Wichita, KS: 0 NHL deaths Tacoma, WA: 0 NHL deaths Whole cohort: 7 NHL deaths Men: 5 NHL deaths Women: 2 NHL deaths

RR for MM (95% CI)

292.e16

Supplementary Table 3 (continued )

Supplementary Table 5 Results of epidemiologic studies of TCDD exposure in communities Author, publication year Level of exposure (with N cases)

RR for NHL (95% CI)

Hardell et al., 1995 [186, 185]

NHL cases: 10.1 (range ¼ <0.4e36) HL case: 2.0 Controls: 2.8 (range ¼ <0.4e6), P ¼ .22 NHL cases: 64.7 (range ¼ 19.9e187) HL case: 14.0 Controls: 29.7 (range ¼ 12.9e53.4), P ¼ .04 NHL cases: mean ¼ 3.2, median ¼ 3.0, range ¼ 1.0e5.8 NR Controls: mean ¼ 3.3, median ¼ 3.2, range ¼ 0.92e9.5; Wilcoxon P ¼ .85 Referent

Mean adipose TCDD concentration (pg/g wet basis) Mean adipose TEQ concentration (pg/g wet basis)

Hardell et al., 1996, 2001 [188, 187]

Adipose TCDD concentration (pg/g lipid basis) (32 NHLs) TCDD grade TCDD grade

3.2, early antigen IgG 80 (7 NHLs, 2 low-grade B cell, 3 highB cell) 3.2, early antigen IgG >80 (6 NHLs, 5 low-grade B cell, 1 highB cell)

TCDD >3.2, early antigen IgG 80 (3 NHLs, 1 low-grade B cell, 2 highgrade B cell) TCDD >3.2, early antigen IgG >80 (6 NHLs, 3 low-grade B cell, 3 highgrade B cell)

TEQ 27.79, early antigen IgG 80 (5 NHLs, 2 low-grade B-cell, 2 highgrade B-cell) TEQ 27.79, early antigen IgG >80 (4 NHLs, 3 low-grade B-cell, 1 highgrade B-cell) TEQ >27.79, early antigen IgG 80 (5 NHLs, 1 low-grade B-cell, 3 highgrade B-cell) TEQ >27.79, early antigen IgG >80 (8 NHLs, 5 low-grade B-cell, 3 highgrade B-cell) Kramárová et al., 1998, 2004 [190, 191]

Recall of spraying missions during war (13 NHLs)

Served in army during war; men only (19 NHLs)

Exposure Opportunity Index > median (57 NHLs)

Exposure Opportunity Index 90% (13 NHLs)

Exposure Opportunity Index tertiles (50 NHLs in top tertile)

Plasmacytoma case: NR

NR

NHL OR ¼ 1.5 (0.32e6.8) low-grade B-NHL OR ¼ 6.0 (0.83e64) high-grade B-NHL OR ¼ 0.59 (0.03e6.3) NHL OR ¼ 0.45 (0.07e2.4) low-grade B-NHL OR ¼ 0.47 (0.02e6.2) high-grade B-NHL OR ¼ 1.1 (0.11e9.8) NHL OR ¼ 0.90 (0.19e4.2) low-grade B-NHL OR ¼ 1.7 (0.19e18) high-grade B-NHL OR ¼ 1.6 (0.19e13) NHL cases: mean ¼ 28, median ¼ 27, range ¼ 7.8e48 Controls: mean ¼ 27, median ¼ 27, range ¼ 11e47; Wilcoxon P ¼ .63 Referent NHL OR ¼ 0.94 (0.17e5.1) low-grade B-NHL OR ¼ 2.1 (0.25e23) high-grade B-NHL OR ¼ 1.5 (0.05e34) NHL OR ¼ 0.75 (0.12e4.7) low-grade B-NHL OR ¼ 0.24 (0.01e3.8) high-grade B-NHL OR ¼ 5.0 (0.34e138) NHL OR ¼ 2.8 (0.52e18) low-grade B-NHL OR ¼ 4.7 (0.56e55) high-grade B-NHL OR ¼ 8.4 (0.57e220) NHL versus all controls: OR ¼ 1.2 (0.57e2.56) NHL versus cancer controls: OR ¼ 0.9 (0.35e2.34) NHL versus non-cancer controls: OR ¼ 1.8 (0.63e5.04) NHL versus all controls: OR ¼ 0.72 (0.36e1.47) NHL versus cancer controls: OR ¼ 1.18 (0.53e2.64) NHL versus non-cancer controls: OR ¼ 0.4 (0.18e1.04) NHL versus all controls: OR ¼ 1.4 (0.89e2.10) NHL versus cancer controls: OR ¼ 1.0 (0.53e1.87) NHL versus non-cancer controls: OR ¼ 2.2 (1.18e4.09) NHL versus all controls: OR ¼ 1.1 (0.52e2.18) NHL versus cancer controls: OR ¼ 1.0 (0.38e2.63) NHL versus non-cancer controls: OR ¼ 1.8 (0.63e5.04) NHL versus all controls: Ptrend ¼ .298 NHL versus cancer controls: Ptrend ¼ .824 NHL versus non-cancer controls: Ptrend ¼ .040

NR

NR

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Adipose TEQ concentration (pg/g lipid basis; 30 NHLs)

RR for HL (95% CI) RR for MM (95% CI)

(continued on next page)

292.e17

Author, publication year Level of exposure (with N cases)

RR for NHL (95% CI)

RR for HL (95% CI) RR for MM (95% CI)

De Roos et al., 2005 [193]

NR

NR

Referent NHL OR ¼ NHL OR ¼ NHL OR ¼ NHL OR ¼

0.77 1.64 2.25 1.35

(0.30e1.98) (0.63e4.30) (0.82e6.20); Ptrend ¼ .06 (1.02e1.79)

Referent NHL OR ¼ NHL OR ¼ NHL OR ¼ NHL OR ¼

0.81 1.28 1.97 1.94

(0.30e2.14) (0.49e3.34) (0.69e5.61); Ptrend ¼ .11 (0.94e4.00)

Referent NHL OR ¼ NHL OR ¼ NHL OR ¼ NHL OR ¼

0.45 (0.13e1.57) 1.38 (0.48e3.94) 1.07 (0.42e2.75); Ptrend ¼ .50 1.002 (0.999e1.005)

NR

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

NR ¼ not reported.

Detectable TCDD (2 NHLs, 5 controls) Total TEQ (pg/g lipid) 15.12 (21 NHLs) >15.12e19.39 (14 NHLs) >19.39e29.46 (30 NHLs) >29.46 (31 NHLs) Per 10 pg per g lipid Dioxin TEQ (pg/g lipid) 4.82 (23 NHLs) >4.82e6.49 (16 NHLs) >6.49e8.92 (26 NHLs) >8.92 (34 NHLs) Per 10 pg per g lipid Total dioxins (mol/g lipid) 0.746 (22 NHLs) >0.746e1.083 (17 NHLs) >1.083e1.888 (35 NHLs) >1.888 (25 NHLs) Per 0.1 mol per g lipid

292.e18

Supplementary Table 5 (continued )

Supplementary Table 6 Design and results of epidemiologic studies of Vietnam veterans without estimated Agent Orange/TCDD exposure Author, publication year

Study subjects

Australian Vietnam veterans Commonwealth 40,030 Male veterans of the Department of Australian Army, Navy, Air Veterans Affairs, Force, and Citizen Military 1998 [147]; Forces and some civilian Australian personnel [211] who served Institute of on Vietnamese land or in Health and Vietnamese waters for 1 d Welfare, between May 23, 1962, and 1999 [148] July 1, 1973 External comparisons with age-standardized expected counts from the Australian Cancer Registry

Wilson et al., 2005 [149]

223 Female veterans of the Australian Army, Navy, Air Force, and Citizen Military Forces and some civilian personnel [211] who served on Vietnamese land or in Vietnamese waters for 1 d between May 23, 1962, and July 1, 1973 External comparisons with age-standardized expected counts from the Australian Cancer Registry 57,864 Male military veterans of the Australian Army, Navy, Air Force, and Citizen Military Forces (excluding civilians; mean age at the first service ¼ 23.9 y) who served in Vietnam between May 23, 1962, and July 1, 1973, and were not deceased before 1982 Comparisons with Australian male national cancer incidence rates standardized by age and calendar year

Definition of outcome

Definition of exposure

Level of exposure (with N cases)

RR for NHL (95% CI)

RR for HL (95% CI)

RR for MM (95% CI)

Active service in Vietnam between May 23, 1962, and July 1, 1973 Cross-sectional health survey in 1996

NHL diagnosed by a doctor since the first day of service in Vietnam, self-reported on a mailed health questionnaire 130 Reported NHL cases contacted for validation study; 99 (76%) responses received; 57 validated based on information from clinicians, National Death Index, National Cancer Statistics Clearing House, Department of Veterans Affairs database, and documentation provided by the veteran; 40 not validated; 9 not able to be validated; 30 with no response (numbers do not add up to 130) See Commonwealth [147]

Active service on Vietnamese land or in Vietnamese waters for 1 d during the Vietnam War (between May 23, 1962, and July 1, 1973)

Vietnam veterans: 137 reported NHLs Vietnam veterans: 62 estimated validated NHLs Expected: 48 NHLs (95% CI, 34e62)

NR NHL prevalence ratio (selfreported) ¼ 2.9 (2.2e4.0) NHL prevalence ratio (estimated validated) ¼ 1.3 (1.0e1.8)

NR

NR

See Commonwealth [147]

Vietnam veterans: 2 NR reported NHLs (95% CI, 0e7) Expected: 0 NHLs (95% CI, 0 e4)

NR

NR

NHL, CLL, HL, and MM incidence ascertained by the Australian Institute of Health and Welfare by matching with the Australian National Cancer Statistics Clearing House; vital status ascertained for 97.5% of cohort (2.5% lost to follow-up)

Active military service in Vietnam during the Vietnam War (May 23, 1962, to July 1, 1973)

Vietnam veterans: 126 NHLs, 58 CLLs, 51 HLs, 31 MMs Expected: 189 NHLs, 37 CLLs, 25 HLs, 47 MMs, excluding unknowns; 193 NHLs, 38 CLLs, 25 HLs, 48 MMs, including unknowns Navy Vietnam veterans: 31 NHLs, 12 CLLs, 7 HLs, 4 MMs Army Vietnam veterans: 86 NHLs, 42 CLLs, 40 HLs, 21 MMs Air Force Vietnam veterans: 9 NHLs, 4 CLLs, 4 HLs, 6 MMs

All veterans HL SIR ¼ 2.05 (1.49e2.61) excluding those with unknown vital status at end All veterans HL SIR ¼ 2.01 (1.45e2.56) including unknowns Navy HL SIR ¼ 1.25 (0.50 e2.57) excluding unknowns Navy HL SIR ¼ 1.22 (0.49 e2.50) including unknowns Army HL SIR ¼ 2.31 (1.59 e3.02) excluding unknowns Army HL SIR ¼ 2.26 (1.56 e2.96) including unknowns Air Force HL SIR ¼ 2.05 (0.56 e5.25) excluding unknowns Air Force HL SIR ¼ 2.01 (0.55 e5.15) including unknowns

All veterans MM SIR ¼ 0.66 (0.43e0.90) excluding those with unknown vital status at end All veterans MM SIR ¼ 0.65 (0.42e0.88) including unknowns Navy MM SIR ¼ 0.40 (0.11 e1.03) excluding unknowns Navy MM SIR ¼ 0.39 (0.11 e1.00) including unknowns Army MM SIR ¼ 0.67 (0.38 e0.96) excluding unknowns Army MM SIR ¼ 0.66 (0.38 e0.94) including unknowns Air Force MM SIR ¼ 1.09 (0.40e2.38) excluding unknowns Air Force MM SIR ¼ 1.06 (0.39e2.31) including unknowns

See Commonwealth [147]

Active service during the Vietnam War Cancer incidence follow-up from 1982 through December 31, 2000 (mean follow-up ¼ 18 y)

All veterans NHL SIR ¼ 0.67 (0.55e0.79), CLL SIR ¼ 1.55 (1.15e1.95) excluding those with unknown vital status at end All veterans NHL SIR ¼ 0.65 (0.54e0.77), CLL SIR ¼ 1.51 (1.12e1.90) including unknowns Navy NHL SIR ¼ 0.76 (0.49 e1.02), CLL SIR ¼ 1.51 (0.78 e2.63) excluding unknowns Navy NHL SIR ¼ 0.74 (0.48 e0.99), CLL SIR ¼ 1.46 (0.75 e2.55) including unknowns Army NHL SIR ¼ 0.67 (0.53 e0.81), CLL SIR ¼ 1.68 (1.18 e2.19) excluding unknowns Army NHL SIR ¼ 0.65 (0.52 e0.79), CLL SIR ¼ 1.65 (1.15 e2.14) including unknowns Air Force NHL SIR ¼ 0.48 (0.22e0.91), CLL SIR ¼ 0.87 (0.24e2.23) excluding unknowns

292.e19

(continued on next page)

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Commonwealth Department of Veterans Affairs, 1998 [140]

Period of exposure and follow-up

292.e20

Supplementary Table 6 (continued ) Author, publication year

Study subjects

Period of exposure and follow-up

Definition of outcome

Definition of exposure

59,179 Male military veterans of the Australian Army, Navy, Air Force, and Citizen Military Forces (excluding civilians; mean age at the first service ¼ 23.9 y) who served in Vietnam between May 23, 1962, and July 1, 1973 Comparisons with Australian male national mortality rates standardized by age and calendar year

Active service during the Vietnam War Mortality follow-up from completion of Vietnam service through December 31, 2001 (w30 to 40 y)

NHL, HL, and MM mortality; vital status ascertained for 97.3% of cohort (2.7% lost to follow-up) from matching to Department of Defense database, Australian State and Territories Registries of Births, Deaths and Marriages, National Death Index, Veterans’ Affairs Client Data Base, Health Insurance Commission Medicare database, National Cancer Statistics Clearing House, Electoral Commission rolls, and Department of Immigration, Multicultural and Indigenous Affairs database; cause of death determined for 97.7% of decedents from Registries of Births, Deaths and Marriages and National Death Index

Wilson et al., 2005 [139]

19,240 Australian male National Service conscripts (selected by an initial, random ballot among all men turning 20 y of age, followed by medical and psychologic assessments) who served in the Australian Army in Vietnam between 1966 and July 1973 (mean age at the end of Vietnam war ¼ 26.3 y), compared with 24,729 Australian male National Service conscripts who served only in Australia (mean age at the end of Vietnam War ¼ 26.1 y) External comparisons with Australian male national mortality and cancer incidence rates standardized by age and calendar year

Active service between 1966 and July 1973 Cancer incidence follow-up from 1982 through December 31, 2000 Mortality follow-up from completion of Vietnam service through December 31, 2001

NHL, CLL, HL, and MM Active military service (by incidence and NHL, HL, and conscription) in Vietnam MM mortality ascertained between 1966 and July 1973 by methods described in [149,150] Vital status ascertained for 96.7% of cohort (2.2% of Vietnam veterans and 4.1% of non-Vietnam veterans lost to follow-up); cause of death determined for 98.9% of decedents

Active military service in Vietnam during the Vietnam War (May 23, 1962, to July 1, 1973)

Vietnam veterans: 70 NHL deaths, 13 HL deaths, 24 MM deaths Expected: 87 NHL deaths, 15 HL deaths, 29 MM deaths excluding unknowns; 90 NHL deaths, 15 HL deaths, 29 MM deaths including unknowns Navy Vietnam veterans: 22 NHL deaths, 2 HL deaths, 3 MM deaths Army Vietnam veterans: 52 NHL deaths, 11 HL deaths, 15 MM deaths Air Force Vietnam veterans: 8 NHL deaths, 0 HL deaths, 6 MM deaths 1963e1979: 8 NHL death, 3 HL deaths, MM NR 1980e1990: 23 NHL deaths, 4 HL deaths, MM NR 1991e2001: 39 NHL deaths, 6 HL deaths, MM NR

Vietnam conscripts: 35 NHLs, 8 CLLs, 12 HLs, 8 MMs Vietnam conscripts: 21 NHL deaths, 4 HL deaths, 5 MM deaths Non-Vietnam conscripts: 39 NHLs, 11 CLLs, 18 HLs, 11 MMs Non-Vietnam conscripts: 19 NHL deaths, 3 HL deaths, 7 MM deaths

RR for NHL (95% CI)

RR for HL (95% CI)

RR for MM (95% CI)

Air Force NHL SIR ¼ 0.47 (0.21e0.89), CLL SIR ¼ 0.85 (0.23e2.17) including unknowns All veterans NHL SMR ¼ 0.78 (0.60e0.96) excluding those with unknown vital status at end All veterans NHL SMR ¼ 0.76 (0.58e0.94) including unknowns Navy NHL SMR ¼ 0.52 (0.25 e0.94) excluding unknowns Navy NHL SMR ¼ 0.50 (0.24 e0.91) including unknowns Army NHL SMR ¼ 0.85 (0.62 e1.08) excluding unknowns Army NHL SMR ¼ 0.83 (0.61 e1.06) including unknowns Air Force NHL SMR ¼ 0.86 (0.37e1.61) excluding unknowns Air Force NHL SMR ¼ 0.84 (0.35e1.57) including unknowns 1963e1979 NHL SMR ¼ 0.85 (0.36e1.59) excluding unknowns 1980e1990 NHL SMR ¼ 1.01 (0.60e1.42) excluding unknowns 1991e2001 NHL SMR ¼ 0.67 (0.46e0.89) excluding unknowns Vietnam versus nonVietnam, incidence: NHL RR ¼ 1.17 (0.72 e1.89), CLL RR ¼ 0.90 (0.31e2.45) Vietnam versus nonVietnam, mortality: NHL RR ¼ 1.42 (0.73 e2.80) All conscripts NHL SIR ¼ 0.67 (0.52e0.82), CLL SIR ¼ 1.24 (0.75e1.94) excluding those with unknown vital status at end NHL SIR ¼ 0.65 (0.50e0.80), CLL SIR ¼ 1.21 (0.73e1.89) including unknowns NHL SMR ¼ 0.87 (0.60e1.14) excluding unknowns NHL SMR ¼ 0.85 (0.59e1.11) including unknowns Vietnam conscripts NHL SIR ¼ 0.72 (0.48e0.95), CLL SIR ¼ 1.18 (0.52e2.33)

All veterans HL SMR ¼ 0.89 (0.46e1.49) excluding those with unknown vital status at end All veterans HL SMR ¼ 0.88 (0.46e1.46) including unknowns Navy HL SMR ¼ 0.58 (0.07 e2.07) excluding unknowns Navy HL SMR ¼ 0.57 (0.07 e2.02) including unknowns Army HL SMR ¼ 1.11 (0.54 e1.94) excluding unknowns Army HL SMR ¼ 1.09 (0.53 e1.91) including unknowns Air Force HL SMR ¼ 0.00 (0.00e2.86) excluding unknowns Air Force HL SMR ¼ 0.00 (0.00e2.81) including unknowns 1963e1979 HL SMR ¼ 0.44 (0.09e1.21) excluding unknowns 1980e1990 HL SMR ¼ 0.87 (0.23e2.19) excluding unknowns 1991e2001 HL SMR ¼ 2.04 (0.74e4.37) excluding unknowns Vietnam versus nonVietnam, incidence: HL RR ¼ 0.90 (0.39 e1.97) Vietnam versus nonVietnam, mortality: HL RR ¼ 1.72 (0.29 e11.77) All conscripts HL SIR ¼ 1.61 (1.03 e2.18) excluding those with unknown vital status at end HL SIR ¼ 1.56 (1.00 e2.12) including unknowns HL SMR ¼ 0.76 (0.30e1.55) excluding unknowns HL SMR ¼ 0.74 (0.29e1.51) including unknowns Vietnam conscripts HL SIR ¼ 1.46 (0.75 e2.55) excluding unknowns HL SIR ¼ 1.43 (0.74 e2.50) including unknowns HL SMR ¼ 0.99

All veterans MM SMR ¼ 0.86 (0.52e1.20) excluding those with unknown vital status at end All veterans MM SMR ¼ 0.83 (0.50e1.16) including unknowns Navy MM SMR ¼ 0.50 (0.10 e1.46) excluding unknowns Navy MM SMR ¼ 0.49 (0.10 e1.41) including unknowns Army MM SMR ¼ 0.81 (0.44 e1.31) excluding unknowns Army MM SMR ¼ 0.79 (0.43 e1.28) including unknowns Air Force MM SMR ¼ 1.70 (0.61e3.61) excluding unknowns Air Force MM SMR ¼ 1.65 (0.59e3.50) including unknowns

Vietnam versus nonVietnam, incidence: MM RR ¼ 2.19 (0.76 e5.98) Vietnam versus nonVietnam, mortality: MM RR ¼ 0.90 (0.22 e3.29) All conscripts MM SIR ¼ 0.91 (0.55e1.43) excluding unknowns MM SIR ¼ 0.89 (0.53e1.39) including unknowns MM SMR ¼ 1.13 (0.58e1.96) excluding unknowns MM SMR ¼ 1.10 (0.56e1.90) including unknowns Vietnam conscripts MM SIR ¼ 0.87 (0.38e1.71) excluding unknowns MM SIR ¼ 0.85 (0.38e1.68) including unknowns MM SMR ¼ 1.08 (0.34e2.46)

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Wilson et al., 2005 [150]

Level of exposure (with N cases)

Vietnam Experience Study O’Brien et al., Male U.S. Army veterans 1991 [141] who first entered military service between January 1, 1965, and December 31, 1971, completed 16 weeks of active service time, held a military occupational specialty other than “duty soldier” or “trainee,” served a single term of enlistment, and were discharged in the enlisted pay grades E-1 to E5, randomly selected from computerized lists, including 9324 Vietnam veterans (mean age at entry into service ¼ 20.3 y) and 8989 Vietnam-era veterans who served in Korea, Germany, or the United States (mean age at entry into service ¼ 20.5 y) For analysis, divided number of NHL cases by sum of veterans who had died and those who had been interviewed (8170 Vietnam veterans and 7564 non-Vietnam veterans), excluding those alive but not interviewed. Expected cases estimated based on U.S. national age-specific cancer incidence data for 1973e1977 Boehmer et al., See [141]; mean age at the 2004 [142] end of study ¼ 53 y

excluding unknowns MM SMR ¼ 1.06 (0.34e2.41) including unknowns Non-Vietnam conscripts MM SIR ¼ 0.95 (0.47e1.70) excluding unknowns MM SIR ¼ 0.92 (0.46e1.64) including unknowns MM SMR ¼ 1.18 (0.47e2.42) excluding unknowns MM SMR ¼ 1.13 (0.45e2.33) including unknowns

(0.26e2.48) excluding unknowns HL SMR ¼ 0.97 (0.26e2.44) including unknowns Non-Vietnam conscripts HL SIR ¼ 1.72 (1.02 e2.72) excluding unknowns HL SIR ¼ 1.66 (0.99 e2.63) including unknowns HL SMR ¼ 0.58 (0.12e1.69) excluding unknowns HL SMR ¼ 0.56 (0.12e1.64) including unknowns

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

excluding unknowns NHL SIR ¼ 0.70 (0.47e0.94), CLL SIR ¼ 1.16 (0.51e2.28) including unknowns NHL SMR ¼ 1.04 (0.60e1.48) excluding unknowns NHL SMR ¼ 1.02 (0.59e1.46) including unknowns Non-Vietnam conscripts NHL SIR ¼ 0.63 (0.44e0.83), CLL SIR ¼ 1.30 (0.65e2.32) excluding unknowns NHL SIR ¼ 0.61 (0.42e0.80), CLL SIR ¼ 1.25 (0.62e2.23) including unknowns NHL SMR ¼ 0.74 (0.44e1.15) excluding unknowns NHL SMR ¼ 0.71 (0.43e1.11) including unknowns NHL prevalence or Active Army service in Vietnam veterans: 4 mortality; prevalent NHL Vietnam between 1965 and prevalent NHLs (1 diffuse ascertained by telephone 1971 mixed lymphocyticinterview in 1985e1986 histiocytic lymphoma, 1 with 87.3% of eligible Burkitt lymphoma, 1 poorly Vietnam veterans and 83.8% differentiated lymphocytic of non-Vietnam veterans; lymphoma, 1 malignant obtained medical records large cell lymphoma), 3 NHL for 41 of 47 veterans with a deaths (1 Burkitt positive report; mortality lymphoma, 1 lymphoblastic ascertained from linkages to lymphoma, 1 reticulum cell the VA Beneficiary sarcoma) Identification Records Expected: 3.9 NHLs Locator Subsystem, Social Non-Vietnam veterans: Security Administration, 0 prevalent NHLs, 1 NHL Internal Revenue Service, death and National Death Index, Expected: 3.5 NHLs and contact with next-ofkin; cause of death coded from death certificate by a nosologist blinded to service location; cause of death independently assigned by a blinded medical review panel based on hospital records, autopsy reports, private physician contacts, coroner or medical examiner records, and law enforcement files

Vietnam versus nonVietnam: P ¼ .07 Vietnam versus expected: P ¼ .10 NHL SIR ¼ 1.8 (0.7e3.7) Non-Vietnam versus expected: P ¼ .14 NHL SIR ¼ 0.3 (0.007e1.6)

NR

NR

Active service in Vietnam between January 1, 1965, and December 31, 1971 Mortality follow-up from

Vietnam veterans: 6 NHL NHL, HL, and MM mortality Active Army service in ascertained from linkages to Vietnam between 1965 and deaths, 2.2 per 100,000 VA Beneficiary 1971 person-years; 2 HL deaths, Identification Records 0.7 per 100,000 person-

NHL crude RR ¼ 0.94 (0.30 e2.93)

NR Crude HL RR ¼ 0.88

NR Crude MM RR ¼ 1.0

(continued on next page)

292.e21

Active service in Vietnam between January 1, 1965, and December 31, 1971 Mortality follow-up from discharge from active duty through December 31, 1983; mean follow-up ¼ 13.7 y for Vietnam veterans, 13.5 y for non-Vietnam veterans

292.e22

Supplementary Table 6 (continued ) Author, publication year

Study subjects

Definition of outcome

discharge from active duty through December 31, 2000; mean follow-up ¼ 29.8 y

Locator Subsystem, Social Security Administration Death Master File, and National Death Index Plus; cause of death obtained (for 98.3% of 1138 newly identified deaths plus 1 of 9 previously identified deaths with missing cause of death) from National Death Index Plus for deaths between 1984 and 2000, and from death certificates for other deaths; unknown vital status assumed to be living

Military service between January 1, 1964, and December 31, 1975 Mortality follow-up from the date of discharge (average year ¼ 1969) through December 31, 1984 (average follow-up ¼ 13 y)

Mortality from lymphosarcoma and reticulosarcoma, other lymphatic tissue cancer, HL, and all lymphoid cancer (calculated from all lymphopoietic cancer minus leukemia and aleukemia), with vital status ascertained from Wisconsin Department of Revenue files, Wisconsin Department of Transportation files, and VA Beneficiary Identification Records Locator Subsystem, excluding those with unknown vital status or missing death certificate (5.6% of Vietnam veterans, 10.1% of non-Vietnam veterans)

Definition of exposure

Level of exposure (with N cases)

RR for NHL (95% CI)

RR for HL (95% CI)

RR for MM (95% CI)

years; 1 MM death, 0.4 per 100,000 person-years Non-Vietnam veterans: 6 NHL deaths, 2.3 per 100,000 person-years; 2 HL deaths, 0.8 per 100,000 personyears; 1 MM death, 0.4 per 100,000 person-years

6 Months of active military service including time in Vietnam between January 1, 1964, and December 31, 1975; Vietnam versus nonVietnam service classified primarily based on the Wisconsin VA DD214 military service separation file

Vietnam veterans: 4 Lymphosarcoma and HL SMRs lymphosarcoma and reticulosarcoma Vietnam veterans: reticulosarcoma deaths, 5 SMRs NR Non-Vietnam other lymphatic tissue Vietnam veterans: veterans versus U.S.: deaths, 4 HL deaths NR 0.62 Non-Vietnam Vietnam-era Non-Vietnam Non-Vietnam veterans: 8 lymphosarcoma veterans versus U.S.: veterans versus and reticulosarcoma deaths, 0.78 Wisconsin: 0.54 6 other lymphatic tissue Non-Vietnam Non-Vietnam deaths, 10 HL deaths veterans versus veterans versus Wisconsin: 0.80 Wisconsin civilians: Non-Vietnam 0.52, P < .05 veterans versus Non-Vietnam Wisconsin civilians: veterans versus 0.74 Wisconsin veterans: Non-Vietnam 0.82 veterans versus Vietnam-era Wisconsin veterans: veterans versus U.S.: 0.96 0.56, P < .05 Vietnam-era Vietnam-era veterans versus U.S.: veterans versus 0.79 Wisconsin: 0.49, Vietnam-era P < .05 veterans versus Vietnam-era Wisconsin: 0.83 veterans versus Vietnam-era Wisconsin civilians: veterans versus 0.47, P < .05 Wisconsin civilians: Vietnam-era 0.77 veterans versus Vietnam-era Wisconsin veterans: veterans versus 0.76 Wisconsin veterans: 0.96 Other lymphoid tissue cancer SMRs Vietnam veterans versus U.S.: 1.02 Vietnam veterans versus Wisconsin: 0.77 Vietnam veterans versus Wisconsin civilians: 1.03 Vietnam veterans versus Wisconsin veterans: 2.89, P < .05 Non-Vietnam

NR

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Other caseecontrol and cohort studies Anderson et al., SMR study: 43,398 1986 [151] Wisconsin Vietnam veterans and 78,940 Wisconsin non-Vietnam veterans (mean age at discharge ¼ 23 y; mean age at the end of follow-up ¼ 37.7 y for Vietnam veterans, 37.9 y for non-Vietnam veterans) drafted in Wisconsin and with permanent residence in Wisconsin, with active service for 180 d between January 1, 1964, and December 31, 1975, and alive at discharge during that period; identified from the Wisconsin VA DD214 military service separation file External comparisons with mortality rates for U.S. overall, Wisconsin overall, Wisconsin civilians, and Wisconsin veterans

Period of exposure and follow-up

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

veterans versus U.S.: 0.57 Non-Vietnam veterans versus Wisconsin: 0.44 Non-Vietnam veterans versus Wisconsin civilians: 0.53 Non-Vietnam veterans versus Wisconsin veterans: 1.16 Vietnam-era veterans versus U.S.: 0.71 Vietnam-era veterans versus Wisconsin: 0.55 Vietnam-era veterans versus Wisconsin civilians: 0.68 Vietnam-era veterans versus Wisconsin veterans: 1.59 All lymphoid cancer SMRs Vietnam veterans versus U.S.: 0.64 Vietnam veterans versus Wisconsin: 0.60 Vietnam veterans versus Wisconsin civilians: 0.65 Vietnam veterans versus Wisconsin veterans: 1.06 Non-Vietnam veterans versus U.S.: 0.60 Non-Vietnam veterans versus Wisconsin: 0.57 Non-Vietnam veterans versus Wisconsin civilians: 0.58 Non-Vietnam veterans versus Wisconsin veterans: 0.94 Vietnam-era veterans versus U.S.: 0.61 Vietnam-era veterans versus Wisconsin: 0.59 Vietnam-era veterans versus Wisconsin civilians: 0.60 Vietnam-era veterans versus Wisconsin veterans: 0.97

292.e23

(continued on next page)

292.e24

Supplementary Table 6 (continued ) Author, publication year Thomas et al., 1991 [143]

McBride et al., 2013 [145]

Period of exposure and follow-up

Definition of outcome

Female U.S. military veterans who served between 1964 and 1972, including 4586 women who served in Vietnam (mean age at entry ¼ 31 y) and 5325 women who served outside the Pacific theater (mean age at entry ¼ 30 y), frequency matched on rank and military occupation; Vietnam veterans identified from Army hospital and administrative support unit morning report records, Air Force computerized personnel file of Vietnam veterans, muster rolls of Navy facilities in Vietnam, and Marine Corps listings of all women assigned to Vietnam; non-Vietnam veterans identified from Army morning report records from units stationed in U.S. and random sample from Air Force, Navy, and Marine Corps lists External comparisons with U.S. female national mortality rates standardized by age, race, and calendar year Random sample of all U.S. Marines on active duty during 1967e1969 (median year of birth ¼ 1946) identified from a computer file, including 10,718 who served in Vietnam and 9346 who did not External comparisons with U.S. male national mortality rates standardized by age, race, and calendar period

Active military service in Vietnam between July 4, 1965, and March 28, 1973 Mortality follow-up from the date of discharge or March 28, 1973, whichever was earlier (average year ¼ 1971 for Vietnam veterans, 1970 for non-Vietnam veterans), through December 31, 1987 (average follow-up ¼ 16.2 y for Vietnam veterans, 17.0 y for non-Vietnam veterans)

Military service between 1967 and 1969 Mortality follow-up from date of discharge from active duty or U.S. military withdrawal from Vietnam (March 28, 1973), whichever was earlier, to date of death or December 31, 1991, excluding those who died while on active duty before March 28, 1973; minimum follow-up w19 y, mean ¼ 21.5 y for Vietnam Marines, 22.5 y for nonVietnam Marines

2783 Male New Zealand veterans with military service in Vietnam between 1964 and 1972, identified from pay records into the VA New Zealand Vietnam veterans’ database Comparisons with New Zealand male national mortality and cancer incidence rates standardized by age and calendar period

Military service in Vietnam between 1964 and 1972 Cancer incidence and mortality follow-up from January 1, 1988, through December 31, 2008

Definition of exposure

Level of exposure (with N cases)

RR for NHL (95% CI)

RR for HL (95% CI)

RR for MM (95% CI)

NHL mortality, with vital Active military service in status ascertained from VA Vietnam between July 4, Beneficiary Identification 1965, and March 28, 1973 Records Locator Subsystem, Social Security Administration, Internal Revenue Service, and National Death Index and military personnel records; cause of death coded by a nosologist blinded to Vietnam service status, based on death certificate obtained from VA claims folder or state vital records department

Vietnam veterans: 3 NHL deaths Non-Vietnam veterans: 3 NHL deaths

Vietnam versus nonVietnam: NHL crude RR ¼ 1.22 Vietnam versus U.S.: NHL SMR ¼ 1.30 (0.26e1.76) Non-Vietnam versus U.S.: NHL SMR ¼ 0.79 (0.16 e2.29)

NR

NR

Lymphosarcoma and reticulosarcoma death, with vital status identified from VA Beneficiary Identification Records Locator System (w94% complete) and cause of death coded by a nosologist blinded to Vietnam service status based on death certificates obtained from VA regional offices, federal records centers, and state vital statistics offices; death certificates missing for 10.4% of Vietnam and 7.5% of non-Vietnam veterans NHL, lymphoid leukemia (described in text as “CLL”), HL, and MM incidence ascertained from linkage to New Zealand Ministry of Health database NHL, HL, and MM mortality ascertained from the Mortality Collection database, which records official underlying cause of death based on the registry of Births, Deaths and Marriages New Zealand,

Military service in Vietnam between 1967 and 1969, identified from military records obtained from the National Personnel Records Center

Vietnam veterans: 4 lymphosarcoma and reticulosarcoma deaths (0.17 per 10,000) Non-Vietnam veterans: 3 lymphosarcoma and reticulosarcoma deaths (0.14 per 10,000)

Vietnam versus nonNR Vietnam: lymphosarcoma and reticulosarcoma SMR ¼ 1.21 (0.27e5.41) Vietnam versus U.S. males: lymphosarcoma and reticulosarcoma SMR ¼ 1.67 (0.45e4.26) Non-Vietnam versus U.S. males: lymphosarcoma and reticulosarcoma SMR ¼ 1.35 (0.27e3.94)

NR

Military service in Vietnam between 1964 and 1972; New Zealand records on Vietnam veterans regarded as “complete”

Vietnam veterans: 14 NHLs, 14 CLLs, 3 HLs, 9 MMs, 3 NHL deaths, 1 HL death, 5 MM deaths Expected: 16.6 NHLs, 7.3 CLLs, 1.4 HLs, 6 MMs, 7.0 NHL deaths, 0.4 HL deaths, 3.2 MM deaths

NHL SIR ¼ 0.85 (0.46e1.42) HL SIR ¼ 2.08 (0.42e6.09) CLL SIR ¼ 1.91 (1.04e3.20) HL SMR ¼ 2.30 (0.03e12.8) NHL SMR ¼ 0.43 (0.09 e1.25)

MM SIR ¼ 1.51 (0.69e2.86) MM SMR ¼ 1.58 (0.51e3.69)

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Watanabe et al., 1995 [144]

Study subjects

Yi et al., 2013 [146]

traffic accident reports, coroners’ inquiries, hospital diagnoses, pathology records, and cancer registry entries 185,265 Korean veterans of Military service in Vietnam NHL (including subtypes), Military service in Vietnam the Vietnam War identified between 1964 and 1973 HL, and MM incidence during Korean involvement by the Ministry of National Cancer incidence follow-up identified from the Korean between 1964 and 1973 Defense and Ministry of from January 1, 1992, National Cancer Incidence Government Administration through December 31, 2003 Database and Home Affairs as of June 2004 Comparisons with Korean male national cancer incidence rates standardized by age and calendar year

All Vietnam veterans: HL All Vietnam veterans: NHL SIR ¼ 1.05 (0.90e1.21) SIR ¼ 0.96 (0.55e1.69) Follicular NHL SIR ¼ 1.09 (0.66e1.79) Diffuse NHL SIR ¼ 1.08 (0.87 e1.33) Diffuse large B-cell lymphoma SIR ¼ 1.14 (0.90 e1.43) T-cell NHL SIR ¼ 1.43 (0.97 e2.12) Peripheral T-cell lymphoma SIR ¼ 1.45 (0.85e2.46) Enlisted soldiers: NHL SIR ¼ 1.00 (0.84e1.19) Noncommissioned officers: NHL SIR ¼ 1.01 (0.70e1.46) Officers: NHL SIR ¼ 2.59 (0.83e8.08) 1992e1997: NHL SIR ¼ 1.01 (0.80e1.29) 1998e2003: NHL SIR ¼ 1.06 (0.89e1.28)

All Vietnam veterans: MM SIR ¼ 1.14 (0.86e1.51)

Military service between July 1, 1958, and April 1, 1973 Death between 1972 and 1983

Mortality as of August 1984 obtained from Massachusetts Department of Public Health

Military service in Vietnam, ascertained from Massachusetts Office of Veterans Services file

Vietnam veterans: 1 death from lymphosarcoma Non-Vietnam veterans: 0 deaths from lymphosarcoma

NR

NR

NR

Military discharge between July 1, 1970, and June 30, 1973 Death between 1970 and 1980

Lymphoma (“lymphoma and Hodgkin disease”) mortality, with vital status ascertained from the VA Beneficiary Identification Records Locator Subsystem and cause of death ascertained from New York State Vital Records and other death certificates

Military service in Vietnam, with information obtained from Defense Manpower Data Center and, for 113 Army personnel discharged in 1971, directly from personnel records

Vietnam veterans: 10 lymphoma deaths Non-Vietnam veterans: 17 lymphoma deaths

Lymphoma mortality OR ¼ 0.96 (0.43e2.18)

NR

NR

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Proportionate mortality and morbidity studies Kogan and Clapp, Deceased Massachusetts 1985 [158] white male veterans (75% aged 20e39 y at death), including 840 who served in Vietnam and 2515 who did not, with eligibility based on service for 6 months between July 1, 1958, and April 1, 1973, residence in Massachusetts for 6 months immediately before entering military service, application for the state military service bonus, and honorable discharge, identified from Massachusetts Office of Veterans Services and linked to mortality files using Social Security numbers Comparison group for standardized mortality ORs: all circulatory diseases except rheumatic heart disease Lawrence et al., 555 Deceased male Vietnam 1985 [154] veterans and 941 deceased male non-Vietnam veterans discharged between July 1, 1970, and June 30, 1973, who died in New York State (excluding New York City) between 1970 and 1980

All Vietnam veterans: 194 NHLs, 17 follicular NHLs, 93 diffuse NHLs, 78 diffuse large B-cell lymphomas, 28 T-cell NHLs, 15 peripheral Tcell lymphomas, 13 HLs, 53 MMs Expected: 185.5 NHLs, 15.6 follicular NHLs, 86.5 diffuse NHLs, 68.6 diffuse large Bcell lymphomas, 19.5 T-cell NHLs, 10.4 peripheral T-cell lymphomas, 13.5 HLs, 46.3 MMs Enlisted soldiers: 137 NHLs Noncommissioned officers: 29 NHLs Officers: 3 NHLs 1992e1997: 72 NHLs 1998e2003: 122 NHLs

292.e25

(continued on next page)

292.e26

Supplementary Table 6 (continued ) Author, publication year Anderson et al., 1986 [151]

Period of exposure and follow-up

Definition of outcome

Definition of exposure

Level of exposure (with N cases)

RR for NHL (95% CI)

PMR study: 923 deceased white male Vietnam veterans and 1571 deceased white male non-Vietnam veterans who served for 6 months between 1964 and 1975 (and 109,225 other deceased white male veterans), were discharged alive from military service and died in Wisconsin between 1964 and July 1983; identified from the Wisconsin VA grave registration file

Military service between January 1, 1964, and December 31, 1975 Death between 1964 and July 1983

Mortality from lymphosarcoma and reticulosarcoma, other lymphatic tissue cancer, HL, and all lymphoid cancer (calculated from all lymphopoietic cancer minus leukemia and aleukemia) ascertained from the Wisconsin Department of Health and Social Services Death Index Tape for 1960 e1978, with back-up from the VA Beneficiary Identification Records Locator Subsystem for vital status

6 Months of active military service including time in Vietnam between January 1, 1964, and December 31, 1975; Vietnam versus nonVietnam service classified primarily based on the Wisconsin VA DD214 military service separation file

Vietnam veterans: 2 Lymphosarcoma and lymphosarcoma and reticulosarcoma reticulosarcoma deaths, 2 PMRs other lymphatic tissue Vietnam veterans deaths, 6 HL deaths versus non-Vietnam Non-Vietnam Vietnam-era veterans: NR veterans: 3 lymphosarcoma Vietnam veterans and reticulosarcoma deaths, versus all other veterans: NR 4 other lymphatic tissue Vietnam veterans deaths, 10 HL deaths versus nonveterans: All other (non-Vietnam-era) NR Wisconsin veterans: 491 Non-Vietnam lymphosarcoma and veterans versus all reticulosarcoma deaths, 576 other veterans: NR other lymphatic tissue Non-Vietnam deaths, 267 HL deaths veterans versus nonveterans: NR Vietnam-era veterans versus all other veterans: 0.32 (0.14e0.73) Vietnam-era veterans versus nonveterans: 0.45 (0.19e1.05) All other veterans versus nonveterans: 0.97 (0.89e1.06) Other lymphatic tissue cancer PMRs Vietnam veterans versus non-Vietnam veterans: NR Vietnam veterans versus all other veterans: NR Vietnam veterans versus nonveterans: NR Non-Vietnam veterans versus all other veterans: NR Non-Vietnam veterans versus nonveterans: NR Vietnam-era veterans versus all other veterans: 0.98 (0.44e2.18) Vietnam-era veterans versus nonveterans: 0.53 (0.24e1.17) All other veterans versus nonveterans: 0.94 (0.87e1.02) All lymphoid cancer PMRs Vietnam veterans versus non-Vietnam veterans: 1.16 Vietnam veterans versus all other veterans: 0.69

RR for HL (95% CI)

RR for MM (95% CI)

HL PMRs Vietnam veterans versus non-Vietnam veterans: 0.97 (0.44 e2.12) Vietnam veterans versus all other veterans: 0.98 (0.44 e2.18) Vietnam veterans versus nonveterans: 0.53 (0.24e1.16) Non-Vietnam veterans versus all other veterans: 0.99 (0.53e1.83) Non-Vietnam veterans versus nonveterans: 0.60 (0.33e1.11) Vietnam-era veterans versus all other veterans: 0.99 (0.61e1.61) Vietnam-era veterans versus nonveterans: 0.58 (0.36e0.93) All other veterans versus nonveterans: 0.97 (0.86e1.10)

NR

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Study subjects

Holmes et al., 1986 [155]

Clapp et al., 1991 [156]

Lymphoma (NHL þ HL) and HL mortality identified from complete listing of West Virginia resident deaths from state Health Statistics Center

Active military service between 1976 and 1973 Death between July 4, 1976, and December 31, 1984

NHL mortality based on Service in U.S. Army I Corps death records obtained from in Vietnam between 1965 VA files and state vital and 1973 statistics offices, excluding combat deaths

Military service during the NHL identified from the Massachusetts Cancer Vietnam era Registry Cancer diagnosis between January 1, 1982, and August 1988

Military service in Vietnam based on state military bonus file, confirmed by review of DD214 discharge form

Military service in Vietnam, that is, identification by the Massachusetts Office of the Commissioner of Veterans Services as having received a state Vietnam bonus (requiring proof of service, usually the DD214 discharge form)

Vietnam veterans: 7 lymphoma deaths, including 5 HL deaths Non-Vietnam veterans: 3 lymphoma deaths, including 1 HL death

Army I Corps veterans: 35 NHL deaths (4 within 0e5 y, 18 within 6e10 y, 13 within 11e15 y, 0 within 16 y after Vietnam tour[s])

Vietnam veterans: 12 NHLs (vs. 183 other cancers, excluding soft tissue sarcoma and kidney cancer) Other Vietnam-era veterans: 36 NHLs (vs. 659 other cancers, excluding soft tissue sarcoma and kidney cancer)

Vietnam veterans versus non-Vietnam veterans: HL PMR ¼ 8.33 (2.71e19.45) Vietnam veterans versus nonveterans: HL PMR ¼ 2.08 (0.69e4.97) Non-Vietnam veterans versus nonveterans: HL PMR ¼ 0.42 (0.01e2.33) All veterans versus nonveterans: HL PMR ¼ 1.28 (0.46e2.76)

Overall: NHL PMR ¼ 0.81 NR (0.56e1.13) 0e5 y after tour(s): NHL PMR ¼ 0.66 6e10 y after tour(s): NHL PMR ¼ 0.89 11e15 y after tour(s): NHL PMR ¼ 0.82 16 y after tour(s): NHL PMR ¼ 0.00 NHL standardized morbidity NR OR ¼ 1.17 (0.58e2.39)

NR

NR

NR

(continued on next page)

292.e27

Massachusetts male veterans diagnosed with cancer between January 1, 1982, and August 1988, including 230 patients (214 aged 30e59 at diagnosis) who received a Vietnam bonus and 939 others (727 aged 30e59 y at diagnosis) who received a Vietnam-era bonus from the State of Massachusetts Comparison group: all eligible patients with cancer other than soft tissue sarcoma, non-Hodgkin lymphoma, kidney cancer,

Active military service between August 1, 1964, and March 28, 1973 Death between 1968 and 1983

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Bullman et al., 1990 [161]

Deceased West Virginia male veterans, including 615 Vietnam and 610 nonVietnam veterans (average age at death ¼ 35.4 and 35.1 y, respectively), who served in the military for 90 d between August 1, 1964, and March 28, 1973, resided in West Virginia for 6 mo before entry into active service and were honorably discharged, identified from a list of applicants for a military bonus offered by the state VA department, linked to state mortality listing by name (1968e1978) or Social Security number (1979 e1983) 6688 Deceased male Army Vietnam veterans who served in I Corps (northernmost part of South Vietnam) between July 4, 1965, and March 1, 1973, and 27,917 deceased male Army non-Vietnam veterans

Vietnam veterans versus nonveterans: 0.52 Non-Vietnam veterans versus all other veterans: 0.73 Non-Vietnam veterans versus nonveterans: 0.55 Vietnam-era veterans versus all other veterans: 0.71 Vietnam-era veterans versus nonveterans: 0.54 All other veterans versus nonveterans: 0.96 NR for NHL (without HL)

292.e28

Supplementary Table 6 (continued ) Author, publication year

Visintainer et al., 1995 [159]

and the cancer of interest (i.e., prostate) Deceased Michigan male veterans (3364 who served in Vietnam and 5229 who served outside Vietnam) identified from the Michigan Department of Management and Budget’s Vietnam-era Bonus List, which required 190 d of honorable service, 6 mo of residence in Michigan, a certified copy of discharge from military service, not being listed as missing in action or deceased during period of service, and application for bonus between December 23, 1974, and June 30, 1980; aged 18e29 y during period of service, aged 35e54 y at death 33,833 Deceased male ground troops who served in the U.S. Army or Marine Corps in Vietnam between July 4, 1965, and March 1, 1973 (age at death 45 y for 24.7% of Army and 16.9% of Marine Corps personnel), and 36,797 who did not serve in Southeast Asia (age at death 45 y for 22.5% of Army and 9.4% of Marine Corps personnel) External comparisons to U.S. male national mortality distribution

Period of exposure and follow-up

Definition of outcome

Definition of exposure

Level of exposure (with N cases)

RR for NHL (95% CI)

RR for HL (95% CI)

RR for MM (95% CI)

Military service between January 1, 1961, and September 1, 1973 Death between 1974 and 1989

NHL and HL mortality, with vital status and cause of death determined by linkage to the Michigan Department of Public Health death certificate database

Military service in Vietnam between January 1, 1961, and September 1, that is, identification as an incountry Vietnam veteran on the Michigan Vietnam-era Bonus List

Vietnam veterans: 32 NHL deaths (29 in non-Blacks), 20 HL deaths (20 in nonBlacks)

All races NHL PMR ¼ 1.51 (1.03e2.13) Non-Black NHL PMR ¼ 1.47 (0.98e2.11)

All races HL PMR ¼ 1.14 (0.70e1.77) Non-Black HL PMR ¼ 1.22 (0.75e1.89)

NR

Active military service between 1965 and 1973 Death between July 4, 1965, and June 30, 1988

NHL (or lymphosarcoma and reticulosarcoma), HL, and MM mortality, with cause of death available for 97.6% of decedents; death certificates obtained from VA regional offices, Federal Records Centers, and state vital statistics offices

Service in U.S. Army or Marine Corps in Vietnam between 1965 and 1973

Army Vietnam veterans: Army Vietnam versus Army Army Vietnam versus Army Army Vietnam versus Army non-Vietnam: MM non-Vietnam: HL 171 NHL deaths (169 non-Vietnam: NHL PMR ¼ 0.86 PMR ¼ 1.01 lymphosarcoma and PMR ¼ 0.88 0e10 y since 0e10 y since reticulosarcoma), 125 HL 0e10 y since leaving Vietnam: MM leaving Vietnam: HL deaths, 36 MM deaths leaving Vietnam: NHL PMR ¼ 0.59 PMR ¼ 0.93 Army non-Vietnam PMR ¼ 0.79 11e15 y since 11e15 y since veterans: 221 11e15 y since leaving Vietnam: MM leaving Vietnam: HL lymphosarcoma and leaving Vietnam: NHL PMR ¼ 0.90 PMR ¼ 1.09 reticulosarcoma deaths, 137 PMR ¼ 0.88 16 y since leaving 16 y since leaving HL deaths, 45 MM deaths 16 y since leaving Vietnam: MM PMR ¼ Vietnam: HL PMR ¼ Marine Corps Vietnam Vietnam: NHL PMR ¼ 1.34 1.12 veterans: 46 NHL deaths (45 1.13 Army Vietnam versus Army Vietnam versus lymphosarcoma and Army Vietnam versus Army þ Marine nonArmy þ Marine nonreticulosarcoma), 25 HL Army þ Marine nonVietnam: MM PMR ¼ Vietnam: HL PMR ¼ deaths, 4 MM deaths Vietnam: NHL PMR ¼ 0.87 1.10 Marine Corps non-Vietnam 0.91 Marine Vietnam Marine Vietnam veterans: 15 Marine Vietnam versus Marine nonversus Marine nonlymphosarcoma and versus Marine nonVietnam: MM PMR ¼ Vietnam: HL PMR ¼ reticulosarcoma deaths, 15 Vietnam: NHL PMR ¼ 0.56 1.85, P < .05 HL deaths, 4 MM deaths 1.68, P < .05 0e10 y since 0e10 y since 0e10 y since leaving Vietnam: NHL leaving Vietnam: MM leaving Vietnam: HL PMR ¼ 1.57 PMR ¼ 0.39 PMR ¼ 1.18 11e15 y since 11e15 y since 11e15 y since leaving Vietnam: NHL leaving Vietnam: MM leaving Vietnam: HL PMR ¼ 1.74, P < .05 PMR ¼ 0.50 PMR ¼ 3.46, P < .05 16 y since leaving 16 y since leaving 16 y since leaving Vietnam: NHL PMR ¼ Vietnam: MM PMR ¼ Vietnam: HL PMR ¼ 1.82 0.79 0.00 Marine Vietnam Marine Vietnam Marine Vietnam versus versus versus Army þ Marine nonArmy þ Marine nonArmy þ Marine nonVietnam: NHL PMR ¼ Vietnam: MM PMR ¼ Vietnam: HL PMR ¼ 0.89 1.10 0.51 Army Vietnam versus Army Vietnam versus Army Vietnam versus U.S. men: HL PMR ¼ U.S. men: lympho/ U.S. men: MM PMR ¼ 0.93 reticulosarcoma 0.92 Army non-Vietnam PMR ¼ 0.93 Army non-Vietnam versus U.S. men: HL Army non-Vietnam versus U.S. men: MM PMR ¼ 0.89 versus U.S. men: PMR ¼ 1.07

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Watanabe et al., 1996 [160]

Study subjects

Clapp et al., 1997 [157]

Military service during the Vietnam era Cancer diagnosis between 1988 and 1993

NR ¼ not reported; PMR ¼ proportionate mortality ratio.

Marine Vietnam versus U.S. men: MM PMR ¼ 0.58 Marine non-Vietnam versus U.S. men: MM PMR ¼ 1.12

MM standardized morbidity OR ¼ 0.7 (0.1e5.7)

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Massachusetts male veterans diagnosed with cancer at ages 35e64 y between 1988 and 1993, including 245 patients who received a Vietnam bonus and 999 others who received a Vietnam-era bonus from the State of Massachusetts Comparison group ¼ all eligible cases with cancer of the gastrointestinal tract (stomach, pancreas, colon, and rectum)

lympho/ Marine Vietnam reticulosarcoma versus U.S. men: HL PMR ¼ 1.08 PMR ¼ 0.73 Marine Vietnam Marine non-Vietnam versus U.S. men: versus U.S. men: HL lympho/ PMR ¼ 0.54, P < .05 reticulosarcoma PMR ¼ 1.06 Marine non-Vietnam versus U.S. men: lympho/ reticulosarcoma PMR ¼ 0.48, P < .05 NHL, HL, and MM identified Military service in Vietnam, Vietnam veterans: 21 NHLs, NHL standardized morbidity HL standardized morbidity OR ¼ 1.5 (0.4e5.5) from the Massachusetts that is, identification by the 5 HLs (vs. 34 OR ¼ 1.6 (0.8e3.3) Massachusetts Office of the gastrointestinal tract Cancer Registry cancers) Commissioner of Veterans Services as having received Other Vietnam-era veterans: 49 NHLs, 10 HLs a state Vietnam bonus (requiring proof of service, (vs. 165 gastrointestinal tract cancers) usually the DD214 All Vietnam-era veterans: discharge form) 10 MMs

292.e29

Study subjects

Period of exposure and follow-up

Definition of outcome

Definition of exposure

Level of exposure (with N cases)

RR for NHL (95% CI)

RR for HL (95% CI)

RR for MM (95% CI)

Revich et al., 2001 [197]

Residents in 1995e1998 of Chapaevsk, Russia, the location of a chemical plant that produced mustard gas and lewisite before 1949, hexachlorocyclohexane (lindane) and derivatives from 1967 to 1987, and liquid chlorine, acids, methyl chloroform, vinyl chloride, and some other chlorinated crop protection chemicals as of 2001 Standardized comparisons with cancer mortality rates in Samara Region, Russia

Cancer mortality in 1995e1998

Lymphoma mortality obtained from official routine medical statistics

Males: 6 lymphoma deaths Females: 9 lymphoma deaths

Lymphoma SMR ¼ 1.1 (0.4e2.3) Lymphoma SMR ¼ 1.1 (0.5e2.1)

NR

NR

Mannes et al., 2005 [195]

Residents in 1972e2001 of 20 Census Collector Districts within 1.5 km of a former chemical manufacturing plant in Rhodes, New South Wales, Australia, that produced creosote, xanthates, dichlorodiphenyltrichloroethane (DDT), 2,4dichlorophenoxyacetic acid, and 2,4,5-trichlorophenoxyacetic acid from 1927 to 1986 Comparisons with cancer incidence and mortality rates in New South Wales and in Ryde and Concord Statistical Local Areas standardized by age, sex, and calendar period Residents in 1970e2001 of New Plymouth, New Zealand, the location of a former 2,4,5trichlorophenoxyacetic acid manufacturing plant from 1962 to 1987 Comparisons with cancer incidence and mortality rates in the remainder of New Zealand standardized by age, sex, and calendar period

Cancer incidence and mortality in 1972e2001

NHL, MM, and lymphoid cancer (calculated from hematopoietic cancers minus leukemia) incidence and mortality obtained from New South Wales Center Cancer Registry

Residence in Chapaevsk based on routine official statistical data TCDD detected in drinking water at <2 pg/L 6e8 km from plant, <2 and 5.0 pg/L at city center in 1998 TCDD detected in human milk at 0.88 ng/L and 23.2 pg/g fat in 40 women from Chapaevsk in 1997 e1998 TCDD detected in serum at 80.5 pg/g lipid in four female workers, 46.4 pg/g lipid in six female residents 1e3 km from plant, and <2.5 pg/g lipid in four female residents 5e8 km from plant in 1998 Residence in 1 of 20 Census Collector Districts within 1.5 km of former plant, based on residential address at the time of cancer diagnosis or death

40 NHLs, 9 MMs, 54 lymphoid cancers (4.9% of 1,106 total cancers)

NR

NR

NR

Cancer incidence and mortality in 1970e2001

NHL, CLL, and HL incidence and mortality obtained from New Zealand Health Information Service’s Mortality Collection and Cancer Registration; cancer incidence data before 1970 e1972 were incomplete because private hospital, death certificate, and incidental autopsy findings were not included

1970e1974: 33 NHLs, 19 NHL deaths, 16 CLLs, 7 CLL deaths, 9 HLs, 7 HL deaths

NHL SIR ¼ 1.75 (1.21e2.46); NHL SMR ¼ 1.56 (0.94e2.44) CLL SIR ¼ 2.51 (1.44e4.08); CLL SMR ¼ 1.69 (0.68e3.49) NHL SIR ¼ 1.30 (0.87e1.87); NHL SMR ¼ 1.58 (1.01e2.35) CLL SIR ¼ 0.87 (0.35e1.80); CLL SMR ¼ 1.75 (0.70e3.61) NHL SIR ¼ 0.83 (0.52e1.26); NHL SMR ¼ 0.96 (0.53e1.62) CLL SIR ¼ 2.55 (1.58e3.91); CLL SMR ¼ 1.37 (0.50e2.98) NHL SIR ¼ 0.71 (0.46e1.06); NHL SMR ¼ 1.31 (0.85e1.95) CLL SIR ¼ 1.43 (0.82e2.34); CLL SMR ¼ 0.84 (0.23e2.17) NHL SIR ¼ 0.81 (0.57e1.14); NHL SMR ¼ 0.94 (0.60e1.42) CLL SIR ¼ 0.90 (0.48e1.55); CLL SMR ¼ 1.13 (0.42e2.50) NHL SIR ¼ 1.07 (0.82e1.38); NHL SMR ¼ 0.69 (0.43e1.06) CLL SIR ¼ 0.88 (0.53e1.39); CLL SMR ¼ 1.33 (0.57e2.64) NHL SIR ¼ 0.79 (0.48e1.25); NHL SMR ¼ 1.02 (0.53e1.79) CLL SIR ¼ 1.07 (0.55e1.87); CLL SMR ¼ 0.78 (0.09e2.84)

HL SIR ¼ 1.19 (0.55e2.27); HL SMR ¼ 1.62 (0.65e3.34)

NR

Read et al., 2007 [196]

Residence within New Plymouth Territorial Authority boundary based on health domicile code for usual place of residence at death or cancer registration or, where unavailable, hospital facility health domicile code (0.87 concordance between health domicile and facility codes for New Plymouth residents, 0.99 for residents of other parts of New Zealand)

1975e1979: 29 NHLs, 24 NHL deaths, 7 CLLs, 7 CLL deaths, 9 HLs, 4 HL deaths 1980e1984: 22 NHLs, 14 NHL deaths, 21 CLLs, 6 CLL deaths, 8 HLs, 6 HL deaths 1985e1989: 24 NHLs, 25 NHL deaths, 16 CLLs, 4 CLL deaths, 9 HLs, 3 HL deaths 1990e1994: 35 NHLs, 23 NHL deaths, 13 CLLs, 6 CLL deaths, 7 HLs, 1 HL death 1995e1999: 61 NHLs, 21 NHL deaths, 19 CLLs, 8 CLL deaths, 4 HLs, 1 HL death 2000e2001: 19 NHLs, 12 NHL deaths, 12 CLLs, 2 CLL deaths, 3 HLs, 0 HL deaths

NR ¼ not reported.

HL SIR ¼ 1.13 (0.52e2.16); HL SMR ¼ 1.16 (0.31e2.99)

HL SIR ¼ 1.06 (0.46e2.10); HL SMR ¼ 2.06 (0.75e4.49)

HL SIR ¼ 1.33 (0.61e2.53); HL SMR ¼ 1.18 (0.24e3.46)

HL SIR ¼ 1.29 (0.52e2.68); HL SMR ¼ 0.61 (0.01e3.45)

HL SIR ¼ 0.65 (0.18e1.68); HL SMR ¼ 0.63 (0.01e3.55)

HL SIR ¼ 1.04 (0.21e3.05); HL SMR ¼ 0

E.T. Chang et al. / Annals of Epidemiology 25 (2015) 275e292

Author, publication year

292.e30

Supplementary Table 7 Design and results of ecologic studies of residential proximity to a TCDD source