Epidemiologic studies of Vietnam veterans potentially provide the most direct evidence of the health effects of Agent Orange exposure. However, because of the small number of highly exposed persons, these studies yield very limited information on cancer. The Vietnam Experience Study (VES), conducted by the Centers for Disease Control (CDC),13 was a historical cohort study that compared 9324 Vietnam Army veterans with 8989 Vietnam-era Army veterans who served elsewhere. It included a mortality study; health interviews; and clinical, psychological, and laboratory evaluation of a random sample of veterans who completed the health interview. The total number of cancer deaths in the VES was inadequate to yield information on specific cancer types. A related effort was the CDC Selected Cancers Study, a population-based case-control study conducted in eight cancer registries that provided data on non-Hodgkin lymphoma,14 sarcomas,15 and other cancers.16 In all of these studies, the number of veterans with substantial exposure to Agent Orange was too small to support firm conclusions.
The Department of Veterans Affairs, formerly the Veterans Administration (VA), also conducted a series of studies beginning in the 1980s. The VA studies ranged from large-scale cohort studies17–19 to case-control studies20 to studies of specific subgroups of veterans.
Both the CDC and the VA studies looked broadly at Vietnam service, without a special focus on Agent Orange exposure (although some VA studies focused on Chemical Corps veterans21,22). In contrast, a third study, the Air Force Health Study, focused specifically on approximately 1200 Ranch Hand veterans directly involved in herbicide distribution and 1300 comparison veterans.23–25 This 20-year prospective study, launched in 1982, involved periodic physical examinations, medical records reviews, and blood dioxin measurements. Information is available at the Air Force Research Laboratory Web site (http://www.brooks.af.mil/AFRL/HED/hedb/afhs/afhs.html). Although this study focused more directly on Agent Orange exposure, the relatively small number of subjects, and the even smaller number with elevated TCDD levels, greatly limited the statistical power to detect increases in cancer incidence.
At the state level, about a dozen states, mostly in the Midwest and Northeast, conducted studies of their veterans, some of which yielded cancer information (eg, New York;26 Massachusetts;27 Wisconsin;28 Michigan29). Finally, a series of studies of Australian Vietnam veterans yielded information on cancer risk.30–33 These studies, too, were limited by small sample sizes, by the absence of detailed exposure assessment, and at least initially by the relatively young age of the veteran populations. As the Vietnam veterans continue to age, additional research should yield additional information about cancer risk.
Because of the limits of the Vietnam veteran studies, indirect sources have provided important information on the potential carcinogenicity of Agent Orange exposure. One of these is data on Vietnamese soldiers and civilians exposed to the same herbicides as United States service personnel, often for more prolonged periods,34–38 although there have been few systematic health studies in these populations. A second indirect source of information is workers exposed to herbicides in other settings, such as herbicide manufacturing workers,39–43 herbicide applicators,44 farmers,45 lumberjacks,46 and forest and soil conservationists,47 who often had much higher serum dioxin levels than Vietnam veterans. Third, people exposed to dioxins after industrial accidents in Germany,48,49 Seveso, Italy,50 and California,51 and after chronic exposures at work52–54 and in the environment55 have been studied. Each of these populations differs from the Vietnam veterans in demographic composition, the nature of the dioxin exposures, and other factors such as diet and concomitant chemical exposures.
Based on this relatively large body of epidemiologic evidence, conclusions can be drawn about several cancers. Each of these is discussed in the paragraphs that follow.
Soft Tissue Sarcoma
Studies of Vietnam veterans have not demonstrated an increase in soft tissue sarcomas. In particular, no association with soft tissue sarcoma was seen in the Ranch Hand study,24 in a study of 10,716 Marines who had served in Vietnam,18 a large case-control study of sarcoma patients in VA hospitals,56 the Selected Cancers Study,15 or studies of veterans in Michigan,29 Massachusetts,27 or other states. A study of Australian Vietnam veterans suggested a large increase in soft tissue sarcomas, but this finding was based on a mail survey of self-reported diagnoses.32 In a follow-up study designed to validate the diagnoses, the excess of soft tissue sarcomas could not be verified.33
However, soft tissue sarcomas have been linked to phenoxy herbicide exposure by a series of case-control studies in Sweden57,58 and by cohort39,40 and case-control41 studies of industrially exposed workers. Many studies of farmers and agricultural workers show an increase in soft tissue sarcomas, which may relate to herbicide exposure. Soft tissue sarcomas have also been linked to dioxin exposure, in a study of 5132 chemical manufacturing workers in the United States,53 in some other occupational studies,59 and in some studies of environmental exposures.60
Most studies of Vietnam veterans have not demonstrated an increase in non-Hodgkin lymphoma (NHL). The Selected Cancers Study showed that Vietnam service was associated with a 50% increased risk of NHL, but self-reported Agent Orange exposure was not associated with increased risk.14 Similarly, in the CDC's Vietnam Experience Study, there were seven NHL deaths among the 8,170 Vietnam veterans and only one NHL death among the 7,564 non-Vietnam veterans. Based on military job titles, there was no suggestion that the seven Vietnam veterans with NHL had sustained Agent Orange exposure.61 In contrast, the Ranch Hand study showed no increase in NHL,62 nor did the VA mortality study of 33,833 Army and Marine Vietnam veterans,19 a case-control study of 201 Vietnam veterans with NHL,63 or numerous state-level studies. A study of Australian Vietnam veterans suggested a large increase in NHL, but this finding was based on a mail survey of self-reported diagnoses.32 In a validation study that attempted to confirm the diagnoses, the number of NHL cases declined to the upper end of the expected range.33
Several case-control studies have found an association between phenoxy herbicide exposure (usually on the job) and NHL.64–67 Numerous other studies of farmers and agricultural workers also suggest this association, although cohort studies of herbicide production workers have generally been negative or report nonsignificant associations based on very small numbers of cases. Dioxin exposure was not associated with NHL in either the NIOSH occupational study54 or the Seveso follow-up,50 although a recent study of a dioxin-exposed area near a municipal solid waste incinerator in France60 suggested a small increase in NHL.
Studies of Vietnam veterans have not demonstrated an increase in Hodgkin disease. In particular, the Ranch Hand study did not show an increase in these tumors,62 nor did a study of 33,833 Army and Marine Vietnam veterans,19 the Selected Cancers Study,16 a case-control study of 283 Vietnam-era veterans with Hodgkin disease,68 or studies of veterans in Michigan,29 New York,26 or other states.
However, Hodgkin disease was linked to phenoxy herbicide and chlorophenol exposure in one case-control study in Sweden,64,69 and another yielded similar results, although without statistical significance.65 Many studies of farmers and agricultural workers show an increase in Hodgkin disease, which may relate to herbicide exposure. The link between Hodgkin disease and dioxin exposure is less clear. The large occupational study of 5,132 chemical manufacturing workers in the United States53,54 did not show an increase in Hodgkin disease. The Seveso follow-up showed no cases of Hodgkin disease in the zone of greatest dioxin exposure, and a small excess of cases in the other zones.50 Other studies have given mixed results.
Studies of Vietnam veterans have not shown a consistent pattern of increases in respiratory cancers (lung, trachea/bronchus, larynx). The VA studies did not reveal increased mortality from these cancers in Vietnam veterans,19,70 nor did the study of Army Chemical Corps veterans.22 The Ranch Hand study suggested an increase in lung cancer, with a relative risk of 3.7, but this finding was based on only 10 deaths, and a high prevalence of smoking in the Ranch Hand population may have accounted for this finding.62 In studies of Australian Vietnam veterans, self-reports suggested an increase in lung cancer (120 cases versus 65 expected),32 but in the validation study, only 46 of these self-reported cases could be confirmed, suggesting a deficit of lung cancer.33
Most studies of workers with occupational herbicide exposure, such as herbicide manufacturing workers,39,40,42,43 herbicide applicators,44 farmers,45 and forest and soil conservationists47 have shown no excess of lung cancer. Similarly, follow-up of the Seveso accident has not shown an association between dioxin exposure and lung cancer,50 although follow-up of industrial accidents in Germany48 and California51 did suggest an increase in respiratory cancers, based on small numbers of cases. Chronic workplace exposures to dioxin have also been associated with increased respiratory cancer, among those with enough exposure to have developed chloracne.54 Together, these data provide little support for the hypothesis that chlorophenoxy acids increase the risk of lung cancer, but they suggest a possible association of dioxin exposure with lung cancer.
While the VA19 and Ranch Hand62 studies did not show an excess of prostate cancer, the Australian veterans study (AIHW) did show an excess, with 212 cases observed and 147 expected.33 Studies of other groups have yielded inconsistent results. Most studies of workers occupationally exposed to phenoxyacetic acid herbicides do not show an excess of prostate cancer. However, there are exceptions. For example, recent studies of pesticide applicators in Florida (exposed to many agents other than herbicides) reported an approximate doubling of prostate cancer incidence and mortality.71,72 Follow-up of the Seveso accident revealed a nonsignificant 20% excess of prostate cancer,50 as did the NIOSH study of chronic dioxin exposure.54 However, follow-up of other acute dioxin exposure incidents48,51 showed no excess of prostate cancer. Overall, the evidence of an association between Agent Orange and prostate cancer is not strong.
None of the studies of Vietnam veterans are informative regarding multiple myeloma risk, because the numbers of cases have been consistently small. However, other studies of people exposed to pesticides, herbicides, and/or dioxins have been suggestive. For example, several studies of farmers and agricultural workers have reported a small increase in multiple myeloma, although other studies show no excess of this neoplasm.73 Follow-up of the Seveso accident shows a deficit of multiple myeloma among exposed males but an excess among females (relative risk 3.7 based on four cases), a disparity that remains unexplained.50 Similarly, the NIOSH study of 5,132 workers exposed to dioxins showed a marginally significant doubling of multiple myeloma risk, based on 10 cases.54 Overall the evidence linking Agent Orange to multiple myeloma is sparse and indirect.
Acute Myelogenous Leukemia in the Children of Veterans
Three studies have pointed to an association between paternal Agent Orange exposure and acute myeloid leukemia (AML) in children. The first was a case-control study of 204 children with AML, reported by the Children's Cancer Study Group, a US-Canada consortium.74 The odds ratio associated with paternal long-term occupational pesticide exposure was 2.7 (95% CI, 1.0–7.0). As for maternal exposure, seven case mothers and no control mothers reported such exposure. The risk was elevated for children diagnosed before the age of six and for children who had sustained direct pesticide exposure. “Pesticides” in this study included both insecticides and herbicides, so it is not clear which agents were associated with the increased risk. The second study was a survey of nearly 50,000 Australian Vietnam veterans.32 This study also found an increase in AML among the children of Vietnam veterans, with a relative risk of 4.3. The risk of acute lymphocytic leukemia (ALL) was not increased in this study.
The third study, a case-control study of 1,805 cases of ALL and 528 cases of AML, was also reported from the Children's Cancer Group.75 Although military service in general conferred no increased risk of childhood leukemia, service in Vietnam or Cambodia was associated with an odds ratio of 1.7 for AML (and no increased risk of ALL). Self-reported exposure to Agent Orange was not associated with increased risk.
Gastrointestinal (GI) Cancer
Cancers of the esophagus, stomach, pancreas, colon, and rectum have been extensively studied in Vietnam veterans, occupational groups with herbicide exposure, and people exposed to dioxins. These studies have yielded a fairly consistent pattern of no association between these exposures and any GI cancer.12 One case-control study in Hanoi suggested that former military service, presumably entailing Agent Orange exposure, was associated with increased risk of hepatocellular carcinoma, but the risk was far smaller than that associated with Hepatitis B infection.76
Similarly, there is a fairly consistent pattern suggesting no association between Vietnam service, occupational herbicide exposure, or dioxin exposure, and brain cancer.12
Available evidence does not permit a conclusion regarding an association between Agent Orange exposure and other cancers, including cancers of the nose and nasopharynx, breast, cervix, uterine corpus, ovaries, liver and biliary tree, bone, kidneys, urinary bladder, testicles, or skin, or leukemia (in veterans themselves, as opposed to their offspring).12
Animal and Laboratory Studies
The chlorophenoxyacetic acid herbicides such as 2,4,5-T and 2,4-D are not considered highly toxic compounds, and high doses are required to cause adverse effects in animals. These compounds have not been associated with cancer in animal bioassays. In vitro laboratory cancer bioassays have also generally been negative, although 2,4-D induced mutations in one bioassay.12
Cacodylic acid is reported to cause lung and bladder tumors, to promote skin cancer in mice, and to be mutagenic in some laboratory tests.77 Picloram has caused increases in benign liver tumors and in thyroid adenomas in rats, but has not been mutagenic in vitro.78
2,3,7,8-TCDD is carcinogenic in animal tests, increasing in a wide variety of tumors in rats, mice, and hamsters. This action is thought to be mediated by the aryl hydrocarbon receptor (AhR), which triggers cellular signaling, DNA binding, and transcriptional activation. In vitro, TCDD does not seem to act as a direct genotoxin but has tumor-promoting activity instead.79
What Expert Agencies Say
Public Law 102-4, the “Agent Orange Act of 1991,” directed the Secretary of Veterans Affairs to request the National Academy of Sciences to review and evaluate the effects of Agent Orange exposure. The Institute of Medicine, part of the National Academy of Sciences, responded by forming the Committee to Review the Health Effects in Vietnam Veterans of Exposure to Herbicides. The Committee has issued a series of studies, beginning with its 1994 Veterans and Agent Orange: Health Effects of Herbicides Used in Vietnam. The IOM reports have assessed the risk of both cancer and noncancer health effects. Each health effect is categorized as having “sufficient evidence of an association,” “limited/suggestive evidence of an association,” “inadequate/insufficient evidence to determine whether an association exists,” or “limited/suggestive evidence of no association.” This framework provides a basis for policy decisions in the face of uncertainty.80 As of the most recent update,12 the associations between Agent Orange exposure and cancer were designated as shown in Table 1. (Note that other diseases, such as chloracne and diabetes, show evidence of an association, but this table shows only the cancers.)
Table TABLE 1. Associations Between Agent Orange Exposure and Cancer
|Sufficient evidence of an association||• Soft-tissue sarcoma|
| ||• Non-Hodgkin lymphoma|
| ||• Hodgkin disease|
| ||• Chronic lymphocytic leukemia (CLL)|
| || |
|Limited/suggestive evidence of an association||• Respiratory cancers (lung, trachea/bronchus, larynx)|
| ||• Prostate cancer|
| ||• Multiple myeloma|
| || |
|Inadequate/insufficient evidence to determine whether an association exists||• Hepatobiliary cancers|
| ||• Nasal/nasopharyngeal cancer|
| ||• Bone cancer|
| ||• Breast cancer|
| ||• Female reproductive cancers (cervical, uterine, ovarian)|
| ||• Urinary bladder cancer|
| ||• Renal cancer|
| ||• Testicular cancer|
| ||• Leukemia (other than CLL)|
| ||• Skin cancers|
| ||• Acute myelogenous leukemia in the children of veterans|
| || |
|Limited/suggestive evidence of no association||• Gastrointestinal cancers (stomach, pancreas, colon, rectum)|
| ||• Brain tumors|
The National Toxicology Program evaluates exposures that may be carcinogenic. Exposures that are thought to be carcinogenic are included in the Reports on Carcinogens, published every two years. Each exposure is assigned to one of two categories: “known to be human carcinogens,” and “reasonably anticipated to be human carcinogens.” The first category includes substances for which human studies (epidemiology studies and/or experimental studies) provide “sufficient evidence” of carcinogenicity in humans. The second category includes substances for which there is limited evidence of carcinogenicity in humans and/or sufficient evidence of carcinogenicity in experimental animals. The National Toxicology Program has not listed the chlorophenoxy herbicides, including Agent Orange, as carcinogens, but 2,3,7,8-TCDD is classified as “known to be a human carcinogen.”81
The International Agency for Research on Cancer (IARC) also evaluates exposures that may be carcinogenic. IARC classifies exposures in one of four categories: Group 1 exposures are those “known to be carcinogenic to humans,” usually based on “sufficient” human evidence, but sometimes based on “sufficient” evidence in experimental animals and “strong” human evidence. Group 2 exposures are divided into two categories. Group 2A (“probably carcinogenic to humans”) has stronger evidence, and Group 2B (“possibly carcinogenic to humans”) has weaker evidence. Group 3 exposures are not considered classifiable, because available evidence is limited or inadequate. Finally, Group 4 exposures are “probably not carcinogenic to humans” based on evidence suggesting lack of carcinogenicity in humans and in experimental animals. IARC has not rated Agent Orange per se, but the chlorophenoxy herbicides, including 2,4-D and 2,4,5-T, are categorized as “possibly carcinogenic to humans” (Group 2B),82 and 2,3,7,8-tetrachlorodibenzo-para-dioxin is categorized as “known to be carcinogenic to humans” (Group 1).79
The Environmental Protection Agency (EPA), through its Integrated Risk Information System, uses a classification scheme very similar to that of IARC. It classifies exposures into one of five categories: (1) human carcinogen, (2) probable human carcinogen, (3) possible human carcinogen, (4) not classifiable as to human carcinogenicity, and (5) evidence of noncarcinogenicity for humans. EPA has not classified either phenoxyacetic acids or TCDD as to carcinogenicity.