Agent or Process	Common Organ or Tissue Sites of Cancer
AMBIENT AND DIETARY EXPOSURE
Acetaldehyde (from alcoholic beverages)	Upper digestive tract
Aflatoxins	Liver
Areca nut	Oral cavity
Aristolochic acid (from plants)	Renal pelvis and ureter
Arsenic and arsenic compounds	Lung, skin
Erionite	Pleura, peritoneum
CULTURAL HABITS
Alcoholic beverages	Oral cavity, pharynx, larynx, esophagus, liver
Betel quid with tobacco	Oral cavity
Tobacco products, smokeless	Oral cavity
Tobacco smoke	Respiratory tract, urinary bladder, renal pelvis, pancreas
Salted fish, Chinese style	Nasopharynx
Solar radiation	Skin
OCCUPATIONAL EXPOSURES
Aluminum production	Lung, urinary bladder
4-Aminobiphenyl	Urinary bladder
Asbestos	Lung, pleura, peritoneum, larynx, gastrointestinal tract
Auramine production	Urinary bladder
Benzene	Leukemia
Benzidine	Urinary bladder
Benzo[a]pyrene	Lung
Beryllium	Lung
Methyl ether	Lung
Boot and shoe manufacture and repair	Nasal sinus
Cadmium	Lung
Chromium (VI) compounds	Lung
Coal combustion (indoors)	Lung
Coal gasification	Lung, urinary bladder, scrotum
Coal-tar pitches	Skin, scrotum, lung
Coal tars	Skin, lung
Coke production	Skin, scrotum, lung, urinary bladder
Dioxin	Multiple sites
Ethylene oxide	Lymphatic, hematopoietic
Fission products	Multiple sites
Formaldehyde	Liver
Furniture and cabinet making	Nasal sinus
Hematite mining	Lung
Iron and steel founding	Lung
Ionizing radiation (all types)	Multiple sites
Isopropyl alcohol manufacture (strong acid process)	Nasal sinus
Leather dust	Nasal sinus
Magenta, manufacture of	Urinary bladder
Mineral oils (untreated and mildly treated)	Skin, scrotum
Mustard gas	Lung, larynx/pharynx
2-Naphthylamine	Urinary bladder
Nickel and nickel compounds	Lung, nasal sinus
Painting	Lung
Polychlorinated biphenyl (PCB-126)	Liver, bile duct, lymphoid tissue
2,3,4,7,8-Pentachlorodibenzofuran	Soft tissue sarcoma, non-Hodgkin lymphoma, cancer of the lung
Rubber industry	Urinary bladder, leukemia
Shale oils	Skin, scrotum
Silica, crystalline	Lung
Soots	Skin, scrotum, lung
Strong inorganic acid mists	Larynx
Talc containing asbestiform fibers	Lung
Toluidine	Urinary bladder
Underground mining with exposure to radon	Lung
Vinyl chloride	Liver, lung, gastrointestinal tract, brain
Wood dust	Nasal cavities, paranasal sinuses
THERAPEUTIC AGENTS
Analgesics mixtures containing phenacetin	Renal, urinary bladder
Azathioprine	Leukemia
N,N-Bis(2-chloroethyl)-2-naphthylamine	Urinary bladder
Busulphan	Lympho-hematopoietic tissue
1,4-Butanediol dimethanesulfonate	Leukemia
Chlorambucil	Leukemia
Chlornaphazine	Urinary bladder
Cyclosporin	Lymphoma
Cyclophosphamide	Urinary bladder, leukemia
Diethylstilbestrol	Breast, cervix
Estrogen replacement therapy	Endometrium, breast
Estrogen, nonsteroidal	Cervix/vagina, breast, endometrium, testes
Estrogens, steroidal	Endometrium, breast
Etoposide	Lymphohematopoietic tissue
Melphalan	Leukemia
8-Methoxypsoralen plus UV radiation	Skin
MOCA (Methylene bis[2-chloroaniline])	Urinary bladder
MOPP combination therapy	Leukemia
Oral contraceptives (combined)	Liver
Oral contraceptives (sequential)	Endometrium
Phenacetin	Renal pelvis, ureter
Semustine (methyl-CCNU)	Leukemia
Sulfur mustard	Lung
Tamoxifen	Endometrium
Thiotepa	Leukemia
Treosulfan	Leukemia
INFECTIOUS AGENTS
Clonorchis sinensis	Liver, bile duct
Epstein-Barr virus	Lymphoma
Helicobacter pylori	Stomach
Hepatitis B virus	Liver
Hepatitis C virus	Liver
Human immunodeficiency virus type 1	Kaposi sarcoma
Human papilloma viruses types 16, 18, others	Cervix
Human T-cell lymphotropic virus type I	Adult T-cell leukemia/lymphoma
Kaposi sarcoma herpesvirus	Kaposi sarcoma
Opisthorchis viverrini	Liver (cholangiocarcinoma)
Schistosoma haematobium	Urinary bladder

Data from the International Agency for Research on Cancer. IARC monographs on the evaluation of carcinogenic risk to humans, vols. 1–104. Lyon (France): IARC; 1970-2012. An updated listing of the overall evaluation of carcinogenicity to humans can be accessed at http://monographs.iarc.fr under the heading “classifications.” (Note: For examples of carcinogenic ionizing radiations, see Table 9-3.)

Chemical carcinogens comprise a diverse array of chemical structures, including both organic and inorganic compounds. Relatively few carcinogens are direct acting, because the innate reactivity of such compounds also tends to make them unstable. Instead, most carcinogens require metabolic activation to reactive species, often in the target cells. Once formed, the reactive intermediates can interact with DNA to produce genetic lesions that can result in mutation of critical cellular genes, including oncogenes and tumor suppressor genes. Metabolic pathways can be influenced strongly by a variety of extrinsic and intrinsic factors and are important determinants of both interindividual and target organ susceptibilities to carcinogens. Carcinogenesis is a dynamic, multistage process through which a normal cell is converted into a malignant one. Although our understanding of the neoplastic process is incomplete, current knowledge provides considerable insight into the critical actions of carcinogens. The goal of this chapter is to highlight the roles of discrete chemical and physical agents in the etiology of human cancers. In turn, fuller understanding of the mechanistic basis for the actions of these carcinogenic agents will allow for more effective means to identify other carcinogens in our environment and to develop preventive strategies to interrupt, block, or reverse the neoplastic process.

Role of Environmental Agents in the Etiology of Human Cancers

The causes of most human cancers remain unidentified; however, considerable evidence suggests that “extraconstitutional” or environmental and lifestyle issues are important contributors. For example, cigarette smoking could be responsible for 25% of all cancers in the United States. The opinion that environmental agents are the principal causes of human cancers is derived largely from the following series of epidemiological observations:

• Although the overall incidence of all cancers is reasonably constant among countries, incidences of specific cancer types can vary up to several hundred fold.

• Large differences in tumor incidence exist within populations of a single country.

• Migrant populations assume the cancer incidence of their new environment within one to two generations.

• Cancer rates within a population can change rapidly.

Although the extent to which environmental agents contribute to human carcinogenesis remains to be defined precisely, a considerable number of epidemiologic studies indicate important roles for the various naturally occurring and manufactured chemicals, radiations, metals, and fibers found in our individual environments. Biological factors such as viruses (e.g., hepatitis B virus and human papilloma virus) and bacteria, elements of our “environment,” are also exceedingly important contributors to the human cancer burden. Biological factors are discussed in Chapter 11.

Chemicals

Polycyclic Aromatic Hydrocarbons

The English surgeon Percival Pott was among the first to document the association of an environmental agent with cancer.² During the late 18th century, he determined that the unusually high incidence of scrotal cancer among chimney sweeps was due to their occupational exposure to soot and tar. As a consequence, recommendations for bathing and use of protective clothing were promulgated by chimney sweepers’ guilds in parts of Europe, but not in England. Subsequent decreases in the incidence of scrotal cancer were observed in continental Europe, demonstrating the efficacy of simple prevention efforts. It was not until the present century that the active carcinogens in soot and coal tar were shown to be polycyclic aromatic hydrocarbons (PAHs).³ This finding came about through the application of coal tar and fractions thereof to the skins of test animals, in which malignant skin tumors subsequently developed. Although many PAHs were identified in coal tar, most of the carcinogenic activity was attributed to the PAH benzo[a]pyrene.

Humans are exposed to PAHs from a variety of sources that include occupation, smoking, diet, and air.⁴ PAHs are readily absorbed into the body through the skin, lungs, and gastrointestinal tract. Occupational and medicinal exposures constitute the highest levels of human PAH exposure (albeit in small groups within the population), whereas diet and smoking are the major sources of exposure to PAHs in the general population. Air concentrations of greater than 10 µg benzo[a]pyrene/m³ are characteristic of topside gas and coke work environments. Broiled, barbecued, or smoked meats and fish contain relatively high concentrations of benzo[a]pyrene (1 to 20 µg/kg).

Cutaneous occupational exposure to PAHs has been associated with increased risk of skin and scrotal cancers in chimney sweeps and in persons exposed to unrefined lubricating oils in the textile and machining industries.¹ Scrotal cancer among mule spinners in the Manchester (United Kingdom) cotton industry was attributed to the saturation of the workers’ trousers with lubricating oil. A review of all admissions for scrotal cancer to the Royal Manchester Infirmary from 1902 to 1922 indicated that 49% had worked as mule spinners, whereas 16% had worked with tar or paraffin. As the textile industry declined in the middle 20th century, an increasing proportion of scrotal cancer was associated with cutting oils used in metal machining.

An excess of lung cancer has been demonstrated among persons with substantial inhalation exposure to PAHs, including roofers and pavers, coke oven workers, certain steel and iron manufacturing workers, and aluminum production workers.⁵ In addition, several studies suggest that workers highly exposed through inhalation also might be at increased risk of cancer at sites other than skin and lung. The strongest evidence for such an association is for bladder cancer, where a dose-response relationship has been demonstrated between PAH exposure and bladder cancer risk in aluminum workers after adjustment for smoking. Other sites with suggestive increases in risk include the pancreas and upper gastrointestinal tract.

Several biochemical pathways are involved in the metabolism of PAHs and of benzo[a]pyrene in particular.³ One major pathway results in highly reactive 7,8-dihydrodiol-9,10-epoxide-benzo[a]pyrene. Experimental studies demonstrate that cultured human lung or colon tissue metabolizes benzo[a]pyrene to reactive metabolites that bind to DNA in cultured tissue. Oral administration of benzo[a]pyrene to rodents produces benzo[a]pyrene-DNA adducts in liver, stomach, colon, and intestine, and cancers of the esophagus, forestomach, intestine, lungs, and mammary gland.

Aromatic Amines

The occurrence of bladder cancer among dye industry workers was reported in 1895 by the German physician Ludwig Rehn, who suggested a causal relationship. With the rapid expansion of the chemical industry during and after World War I, increased risk of bladder cancer was observed among workers employed in chemical manufacturing and textile dyeing.⁶ An industry-wide study of workers exposed to dyes in England and Wales demonstrated increased risks of bladder cancer among men exposed to 1-naphthylamine (observed [O]/expected [E] = 8.6), benzidine (O/E = 13.9), 2-naphthylamine (O/E = 86.7), or mixed dyes (O/E = 54.7). IARC subsequently considered that the cancer hazard associated with exposure to 1-naphthylamine was due to the probable contamination of commercial-grade 1-naphthylamine with 4% to 10% 2-naphthylamine.⁷ Additional studies in the dye industry identified auramine O and magenta as human bladder carcinogens. Increased risk of bladder cancer among rubber workers and in the electric cable industry has been attributed to the naphthylamine added to rubber as an antioxidant.

Excess risk of bladder cancer has been observed in the silk-dyeing industry, in which benzidine-based dyes are used extensively. An elevated incidence of bladder cancer associated with benzidine manufacturing and production in the United States and Japan is complicated by probable co-exposure to 2-naphthylamine and o-toluidine. The production of these aromatic amines has declined in recent years; the result has been a considerable reduction in bladder cancer among workers in these industries.

Aromatic amines are metabolized and excreted through a process involving acetylation by N-acetyltransferase.⁸ Genetic variation in one of the genes, NAT2, encoding this enzyme produces either rapid or slow metabolic phenotypes in humans. Analysis of the NAT2 phenotypes of patients with bladder cancer from the dye industry indicates that persons showing the slow phenotype could be more susceptible to bladder cancer caused by aromatic amines. This finding is consistent with a meta-analysis indicating that NAT2 slow acetylation status is associated with an increased risk of bladder cancer in the general population.⁹

Benzene

Exposure to benzene was suspected to be the cause of leukemia in a number of individual cases and case series reported worldwide between 1928 and 1976.¹ Case-control studies indicated increased risks of nonlymphocytic leukemia among workers in Sweden exposed to petroleum products containing benzene and of lymphomas among workers in New York State who were exposed to benzene. Prospective studies conducted in the rubber industry provide the most convincing evidence for an association between benzene exposure and leukemia. Most of the excess leukemia in this industry is found among rubber workers exposed to solvents, including benzene. Excess mortality from leukemia has been observed among former employees of a rubber film production plant (O/E = 4.7) and a rubber coating plant (O/E = 3.7).

Aflatoxins

The hepatotoxic effect of aflatoxins was first recognized when aflatoxin-contaminated feed was inadvertently fed to poultry. Subsequent animal studies demonstrated the carcinogenic potential of the aflatoxins, particularly aflatoxin B₁. The aflatoxins are produced by the fungal strains Aspergillus flavus and A. parasiticus. Grains and foodstuffs for human consumption such as corn, peanuts, and rice can become contaminated with aflatoxin during growth or storage. The considerable variation in levels of human exposure to aflatoxin worldwide is determined by climate and by the preventive measures used to protect susceptible foods from mold contamination and growth.¹⁰

Dietary aflatoxin is correlated with high liver cancer rates in sub-Saharan Africa and Asia. Case-control studies in the Philippines and Mozambique show an increased risk of liver cancer with estimated levels of aflatoxin consumption. The co-carcinogenic role of hepatitis B virus (HBV) infection and dietary aflatoxin in liver cancer has been the focus of several studies. The incidence of liver cancer in different regions of Swaziland correlated more closely with aflatoxin intake than with HBV infection. A prospective study conducted in Guangxi Province, China, compared the incidence of liver cancer in regions of high and low aflatoxin contamination and determined HBV infection status.¹⁰ A strong interaction between aflatoxin exposure and HBV-positive status was observed for relative risk of liver cancer. Among HBV-positive individuals, the incidence of liver cancer was 649 per 100,000 in the high-aflatoxin region and 66 per 100,000 in the low-aflatoxin region, whereas among HBV-negative individuals, the incidence of liver cancer was 99 per 100,000 and <1 per 100,000 in high- or low-aflatoxin regions, respectively.

The new techniques of molecular dosimetry for human carcinogen exposure have been applied in populations exposed to aflatoxin. With individual exposures often in excess of 10 to 100 µg/day, the presence of aflatoxin metabolites and DNA adducts can be quantified in the urine after exposure. The association of urinary aflatoxin-DNA adducts with risk of liver cancer also has been demonstrated in a prospective epidemiological study.¹¹

Tobacco Chemicals

Tobacco use causes more cancer deaths worldwide than any other human activity. Cigarette smoking is associated with cancers of the lung, oral cavity, pharynx, larynx, esophagus, bladder, renal pelvis, and pancreas (Box 9-1). The use of smokeless tobacco (chewing tobacco or snuff) leads to cancer of the oral cavity. Thus although combustion enhances the carcinogenic properties of tobacco, it is not required for cancer induction.

Box 9-1 Health Effects and Control of Smoking

A series of reports from the U.S. Surgeon General since 1964 have argued that cigarette smoking is the most significant source of preventable morbidity and premature mortality in high-income countries. An estimated annual excess mortality of 400,000 is attributed to cigarette smoking in the United States. These deaths are the result of coronary heart disease, cancer, and various respiratory diseases.⁴⁰ Cancers associated with smoking or smokeless tobacco use include those of the lung, oral cavity, esophagus, pharynx, larynx, and bladder. Weaker associations have been reported for cancers of the pancreas, kidney, stomach, nasopharynx, and cervix.¹² The overall increase in risk of disease among smokers compared with nonsmokers is about tenfold for lung cancer, sixfold for chronic obstructive pulmonary disease, and twofold for myocardial infarction. The combined effect of smoking-related diseases on the average life expectancy of smokers is a reduction of 5 to 8 years. On a worldwide basis, the data are equally disturbing. An estimated 6 million deaths per year are attributed to tobacco use, with upward of 80% of these deaths in low- and middle-income countries. If current use trends continue unabated, estimates as high as 8 million deaths due to tobacco use annually could be expected by the year 2030, and a total of 1 billion this century.⁴¹

The addictive properties of tobacco smoking, due primarily to its nicotine content, cause both physiological and psychological dependence. Withdrawal symptoms can be severe and include irritability, aggressiveness, hostility, depression, difficulty concentrating, and a craving for tobacco. These pharmacologic factors are reinforced by social factors such as peer pressure, emulation of family role models, and cultural influences. Because most smokers begin smoking, and often form lifelong smoking habits, in their teenage years, preventive educational measures should be focused on this age group.

Smoking control measures use various strategies: cessation programs, clinical or community interventions, governmental or private-sector regulations, taxation of tobacco products, warning labels, and smoking prevention programs. Although the majority of ex-smokers have achieved abstinence without extensive personal assistance from organized cessation programs, other control measures (e.g., national health education programs, physician counseling, and indoor smoking regulations) and family pressure are important influences contributing to smoking cessation. Smoking prevention programs, particularly in schools, and government taxation have been somewhat effective in reducing the initiation of smoking among children and adolescents in developing countries. These approaches will need to be applied more vigorously, especially in low- to middle-income countries, to stem the expansion of smoking worldwide. New challenges also are arising with the creation of new and emerging tobacco products such as dissolvable tobacco products, e-cigarettes, and hookah tobacco. The recent Family Smoking Prevention and Tobacco Control Act (2009) in the United States has given the U.S. Food and Drug Administration broad authority to regulate the manufacturing, distribution, and marketing of these and more traditional tobacco products to protect public health.

Although the carcinogenic properties of tobacco tar were first demonstrated experimentally during the 1920s, evidence of a human cancer risk from the use of tobacco did not appear until 1939, when Muller and colleagues ¹² reported an association between tobacco use and lung carcinoma in Germany. Subsequent epidemiological studies conducted in the United States and the United Kingdom during the next decade confirmed this causal relationship. These findings met with considerable resistance in both the scientific community and the general public, however. Unlike occupational exposure to carcinogens, which was subject to regulation in many countries, tobacco exposure was a personal habit considered by many users to be more pleasurable than dangerous. Unfortunately, the addictive characteristics of nicotine, a major constituent of tobacco, made it more difficult for tobacco users to reduce their consumption. Societal acceptance of a causal association with lung cancer was advanced by the first reports from the Royal College of Physicians in the United Kingdom (1962) and from the Surgeon General in the United States (1964) regarding the risks of tobacco use. By contrast, the tobacco industry has steadfastly resisted attempts to educate the public to the health hazards of tobacco use and has continued to market cigarettes aggressively, particularly in developing countries. Explosive increases in the incidence of lung cancer, probably even outranking those already occurring in the United States, can be anticipated in these countries during the next few decades. Based on tobacco use trends, it is estimated that more than 1 million new cases per year of lung cancer will occur in China in the 21st century.¹²

More than 3000 chemicals have been identified in cigarette smoke, of which at least 30 are known to be carcinogenic in animals (Table 9-2). The gas phase of tobacco smoke contains several carcinogenic or tumor-promoting compounds, including dimethylnitrosamine, dialkylnitrosamines, vinyl chloride, acrolein, and benzene. The particulate phase contains carcinogenic and co-carcinogenic PAHs, methylated PAHs, heterocyclic hydrocarbons, chlorinated hydrocarbons, phenols, catechols, and metals. Organ-specific carcinogens in the particulate phase include N-nitrosamines (and precursors), which have been associated with esophageal and pancreatic cancers, and aromatic amines, which are associated with kidney and bladder cancers. An important finding from experimental studies is the strong interactive effect observed when certain mixtures of these compounds are assayed for carcinogenic potential.

Table 9-2

Tumorigenic Agents in Tobacco Smoke

Compounds	Mainstream Smoke Compounds (per Cigarette)
POLYCYCLIC AROMATIC HYDROCARBONS
Benzo[a]anthracene	20-70 ng
Benzo[b]fluoranthene	4-22 ng
Benzo[f]fluoranthene	6-21 ng
Benzo[k]fluoranthene	6-12 ng
Benzo[a]pyrene	20-40 ng
Chrysene	40-60 ng
Dibenz[a,h]anthracene	4 ng
Dibenzo[a,i]pyrene	1.7-3.2 ng
Dibenzo[a-1]pyrene	Detectable
Indenol[1,2,3-c,d]pyrene	4-20 ng
5-Methylchrysene	0.6 ng
AZA-ARENES
Quinoline	1-2 µg
Dibenz[a,h]acridine	0.1 µg
Dibenz[a,j]acridine	3-10 ng
7H-Dibenzo[c,g]carbazole	0.7 ng
N-NITROSAMINES
N-Nitrosodimethylamine	0.1-180 ng
N-Nitrosethylmethylamine	3-13 ng
N-Nitrosodiethylamine	0-25 ng
N-Nitrosopyrrolidine	1.5-110 ng
N-Nitrosodiethanolamine	0-36 ng
N-Nitrosonornicotine	0.12-2.7 µg
4-(Methylnitrosamine)-1-(3-pyridyl)-1-butanone	0.08-0.77 µg
N-Nitrosoanabasine	0.14-4.6 µg
AROMATIC AMINES
2-Toluidine	30-200 ng
2-Naphthylamine	1-22 ng
4-Aminobiphenyl	2-5 ng
ALDEHYDES
Formaldehyde	70-100 µg
Acetaldehyde	18-1400 ng
Crotonaldehyde	10-20 µg
MISCELLANEOUS ORGANIC COMPOUNDS
Benzene	12-48 µg
Acrylonitrile	3.2-15 µg
2-Nitropropane	0.73-1.21 µg
Ethylcarbamate	20-38 ng
Vinyl chloride	1-16 ng
INORGANIC COMPOUNDS
Hydrazine	24-43 ng
Arsenic	40-120 ng
Nickel	0-600 ng
Chromium	4-70 ng
Cadmium	41-62 ng
Polonium-210	0.03-1.0 pCi

From U.S. Department of Health and Human Services. Reducing the health consequences of smoking: 25 years of progress. Washington, DC: U.S. Department of Health and Human Services; 1989.

Secondhand smoke contains many of the same carcinogens that are inhaled by smokers. It is now known that secondhand smoke causes lung cancer in adults who do not themselves smoke.¹³ Nonsmokers exposed to secondhand smoke at home or work are at a 20% to 30% increased risk for the development of lung cancer. It is estimated that secondhand smoke causes an additional 3400 lung cancer deaths each year in nonsmokers in the United States.¹⁴

Chemotherapeutic Agents

The systemic toxicity of sulfur mustard gas among soldiers exposed during World War I led to investigations of the mechanism of action of nitrogen mustard compounds. The cytotoxic effect observed in lymphatic tissues was subsequently replicated and studied in experimental animal models. This property of nitrogen mustards and other alkylating compounds prompted their use as antineoplastic drugs during the 1940s. Several other types of drugs were developed at that time for use in the treatment of cancer, including the antibiotic actinomycin A and the antimetabolite methotrexate. In later decades a variety of alkylating agents were introduced (e.g., chlorambucil, cyclophosphamide, bis-chloroethylnitrosourea, busulfan, and cisplatin), along with antimetabolites (e.g., 5-fluorouracil and 6-mercaptopurine), antibiotics (e.g., Adriamycin, bleomycin, and daunomycin), and mitotic inhibitors (e.g., vincristine and vinblastine) as antineoplastics.

As early as 1948, the carcinogenic properties of the anticancer drug 4-aminostilbene and its metabolites were reported. This and subsequent findings led to the institution of carcinogenesis bioassays for new anticancer drugs under development by the National Cancer Institute, which is now part of the standard toxicologic assessment. Preneoplastic dysplasias were observed frequently in the epithelial tissues of patients with cancer who were undergoing chemotherapy. The appearance of frank second malignancies among patients treated by chemotherapy was reported during the 1970s. In addition, renal transplant patients receiving anticancer drugs for immunosuppression showed excess risks of mesenchymal and epithelial cancers.

The successful treatment of Hodgkin disease with multiagent chemotherapy is associated with the long-term complication of acute myeloid leukemia (AML) and non-Hodgkin lymphoma. Increased risk of AML among patients treated for non-Hodgkin lymphoma, ovarian cancer, multiple myeloma, or small cell carcinoma of the lung also has been attributed to antineoplastic therapy. The risk of AML is most strongly associated with the alkylating antineoplastics—particularly cyclophosphamide, melphalan, busulfan, treosulfan, and semustine (methyl-CCNU)—or with combination chemotherapies that include alkylating agents.

A large case-control study conducted in collaboration with 11 population-based cancer registries and two large oncology hospitals in Europe and Canada identified 114 cases of leukemia among 99,113 ovarian cancer survivors.¹⁵ Patients receiving chemotherapy alone had a relative risk of 12.0 for leukemia compared with patients treated with surgery alone. In contrast, patients receiving radiation therapy alone (compared with surgery alone) had no significant increase in the risk of leukemia. In order of decreasing leukemogenic potency, the drugs melphalan, thiotepa, chlorambucil, cyclophosphamide, and treosulfan were independently associated with significantly increased risk of leukemia. Combination treatment with Adriamycin and cisplatinum also increased the risk of leukemia, indicating that one or both of these drugs is leukemogenic in humans.

Radiation

Ultraviolet

Solar ultraviolet (UV) radiation is the major physical carcinogen in our environment and the primary cause of skin cancer in humans. New basal cell carcinoma or squamous cell carcinoma of the skin will develop in more than 800,000 persons each year in the United States, making nonmelanoma skin cancer the most common cancer.⁷ The incidence of both nonmelanoma and melanoma skin cancer among light-skinned individuals is increasing at a rate of 3% to 5% per year in the United States. This increase has been attributed to changing lifestyle and leisure habits during the past five decades, resulting in an increase in the number of people receiving greater exposure to sunlight. Although basal cell carcinoma occurs more frequently than squamous cell carcinoma (the ratio of basal cell to squamous cell carcinoma is approximately 3 : 1), the incidence of squamous cell carcinoma seems to be increasing more rapidly than that of basal cell carcinoma. In addition, squamous cell carcinoma metastasizes more frequently and is responsible for more deaths than are caused by basal cell carcinoma.

The relationship between solar UV exposure and skin cancer in humans has been demonstrated from incidence data in human populations residing at different latitudes. The incidence of nonmelanoma skin cancer shows a generally increasing trend with decreasing latitude among persons with similar skin types.⁷ Nonmelanoma skin cancers and the premalignant skin neoplasm, actinic keratosis, also are associated with cumulative lifetime UV exposure estimated from outdoor activities. This association is particularly apparent among persons with outdoor occupations such as farmers and sailors. The anatomic distribution of these skin neoplasms, primarily on sun-exposed areas, including the face, ears, neck, and hands, is consistent with a solar etiology. The phenotypic characteristics of light skin complexion, ease of sun burning (skin type), and light hair color are known to enhance the risk of nonmelanoma skin cancer. Pigmentation of the skin, either constitutive or induced (as in tanning), clearly plays an important role in protecting skin from the carcinogenic effects of UV radiation. Persons with moderately to heavily pigmented skin (e.g., Latinos, Hispanics, and black persons) have much lower rates of skin cancer than do persons with poorly pigmented or nonpigmented skin (e.g., Celts and albinos). The importance of pigmentation is also demonstrated by the finding that susceptibility to sunburn is a strong indicator of risk of both basal and squamous cell carcinoma.

Molecular evidence also supports an etiologic role of solar UV in human skin cancer. Increased levels of DNA photodamage are detected in the normal epidermis of individuals after exposure to solar UV radiation. In addition, analysis of mutational spectra in human nonmelanoma skin cancer DNA shows that mutations specific for UV radiation (dipyrimidine mutations) often are present.

Animal studies confirm the carcinogenic effects of UV radiation and indicate that the UV-B portion (280 to 320 nm) of the solar spectrum is primarily responsible for the carcinogenic properties of sunlight. This wave band encompasses the long-wavelength end of the absorbance spectrum of DNA and has been shown to cause mutations in mammalian cells. Stratospheric ozone efficiently absorbs UV-B wavelengths below 300 nm, thereby determining the short-wavelength end of the solar spectrum reaching the earth’s surface. Concern over the destruction of stratospheric ozone due to environmental pollution with chlorofluorocarbons, resulting in increased intensity of UV-B radiation at the earth’s surface, has encouraged the refinement of risk estimates for human skin cancer under conditions of reduced atmospheric ozone.

Ionizing

The discovery and manipulation of ionizing radiation in the early 20th century led to detrimental health effects among many researchers. Toxicity, radiation burns, and cancer were observed among handlers of radioactive materials. The deaths of Marie Curie and Thomas Edison’s assistant from cancer have been attributed to severe radiation exposure. The use of radium in luminous paint during the 1930s led to a high incidence of osteosarcoma among dial painters who inadvertently ingested radium when shaping their brush tips with the tongue.¹ By the 1940s, an elevated incidence of leukemia was observed among radiologists. After World War II, excessive rates of leukemia were observed among atomic bomb survivors and among patients treated with x-rays.

Epidemiological studies of populations exposed to high doses of radiation indicate increased risks for a variety of cancers, depending on the type of radiation and route of exposure (Table 9-3). Among atomic bomb blast survivors, excessive rates of leukemias appeared within several years of exposure, whereas excessive rates of cancers of the breast, lung, esophagus, thyroid, colon, bladder, and ovary, as well as multiple myeloma, appeared only 20 to 25 years later. In contrast, populations exposed to nuclear weapons fallout show only an excessive risk of thyroid cancer due to radioactive iodine. Heavy exposure to x-rays for diagnostic or therapeutic procedures has been associated with increased risk of the following types of cancers: leukemia after in utero exposure; breast cancer after repeated chest exposure; leukemia, lung, stomach, and esophagus after spinal exposure; and thyroid, skin, and neck after scalp or thymus exposure.⁵

Table 9-3

Examples of Radiation-Induced Cancers

Sources of Exposure	Exposure Circumstances	Cancer Types
EXPLOSIONS OF NUCLEAR WEAPONS
Blast	Atomic bombing survivors in Hiroshima and Nagasaki	Leukemia, breast, lung, thyroid, stomach, colon, multiple myeloma, esophagus, ovary
Fallout	Populations exposed through atmospheric testing, including Marshall Islanders, veterans in the Pacific, general population in Nevada, Utah	Thyroid
DIAGNOSTIC PROCEDURES
X-ray	Children exposed in utero	Leukemia
Thorotrast	Cerebral and limb angiography; of biliary passages	Liver
Fluoroscopic x-ray	Monitoring of lung infections in patients with tuberculosis	Breast
THERAPEUTIC PROCEDURES
X-ray	Postpartum mastitis	Breast
X-ray	Ankylosing spondylitis	Leukemia, lung, stomach, esophagus
Cobalt-60	Treatment for cancer of the cervix	Leukemia, stomach, rectum, bladder, vagina, female genital, lung, buccal cavity, nasopharynx, esophagus
X-ray	Treatment of benign head and neck conditions	Thyroid, skin, central nervous system
Radium-224	Ankylosing spondylitis, bone tuberculosis	Bone sarcoma
PROFESSIONAL EXPOSURES
X-ray	Early radiologists	Skin, leukemia
Radon	Uranium, hard-rock miners	Lung cancer
X-ray, γ-ray, neutrons	Nuclear industry	Multiple myeloma
Radium isotopes	Radium dial painters	Bone, head sarcoma

Adapted from Higginson J, Muir CS, Munoz N. Human cancer: epidemiology and environmental causes. Cambridge monographs on cancer research. Cambridge: Cambridge University Press; 1992.

The use of cobalt-60 x-ray treatment for cervical cancer is associated with leukemia and cancers of the stomach, rectum, bladder, vagina, buccal cavity, nasopharynx, and lung. Because most radiation exposure in ambient or occupational environments occurs as protracted low-dose exposure, an important public health concern is the cancer risk from low-level exposure. However, most risk estimates are extrapolated from high to low doses and from acute to chronic exposures and therefore are subject to several assumptions that profoundly influence the resulting low-level risk estimates. Populations with potential (or known) low-level radiation exposures include employees in the nuclear industry, individuals living near nuclear production or storage facilities, military personnel participating in atmospheric nuclear weapons tests or living near test sites, patients receiving diagnostic radiation, and residents of buildings with radon contamination. In addition, nuclear accidents such as the Chernobyl incident in the former Soviet Union produce both acute and chronic exposure to local and distant populations.

Radon

Radon gas is encountered in hard rock mining for iron, tin, fluorspar, and uranium.⁷ The radioactive decay of radon and its products produces alpha particles. The earliest reports defining an association between lung cancer and mining described the high rate of lung cancers among uranium miners in the Schneeberg region of Czechoslovakia in the late 19th century. Many potential causes were proposed, including radon inhalation. Studies of lung cancer mortality among Colorado uranium miners demonstrated dose-related increases in lung cancer risk in miners with protracted exposure to radon.¹ In addition, small cell undifferentiated carcinomas predominated in miners with high exposure to radon, in contrast to the typical distribution of pulmonary cancer pathology in the general U.S. population. An elevated risk of lung cancer also has been reported for iron ore miners in England, France, and Sweden; however, the proportion of risk attributable to radon in these populations is more difficult to assess.

Analysis of lung cancer mortality and smoking in Colorado uranium miners suggests a greater than additive mortality rate for cumulative radon exposure and cumulative cigarette smoking.⁷ In other words, the increased risk of lung cancer among miners compared with nonminers is larger when comparing smokers than when comparing nonsmokers. Interestingly, among atomic bomb survivors, cigarette smoking and radiation exposure only have an additive effect for lung cancer risk. This anomaly has been attributed to the different exposure patterns experienced by atomic bomb survivors (acute) and uranium miners (protracted).

Metals

Arsenic

Medicinal use of inorganic arsenic was associated with skin cancers in the early 20th century. More recently, excessive rates of skin cancer have been observed in populations exposed to drinking water contaminated with arsenic, whereas excessive rates of lung cancer have been found in populations with occupational exposure to inorganic arsenic compounds.¹ An increased risk of lung cancer of six- to fourteenfold was reported for gold miners in Rhodesia, where the ore contains arsenic. Chronic arsenism also was prevalent among these miners. Several studies in Japan, Sweden, and the United States have documented excessive rates of lung cancers among workers involved in copper smelting. Inorganic arsenic is a byproduct of the smelting process and also is used as a hardener. Two large retrospective studies of copper smelters have shown that lung cancer mortality is related to estimated arsenic exposure.

Another source of occupational exposure is the manufacture and use of arsenical pesticides.¹ An early study of mortality among workers at a factory manufacturing arsenical sheep dip in Wales found an excess of skin and lung cancers, particularly among persons directly involved in the chemical processes. Case-control studies in two U.S. plants manufacturing arsenical pesticides found arsenic dose-related increases in the risk of lung cancer. Reports of skin and lung cancers among vineyard workers with exposure to arsenic fungicides and pesticides appeared during the late 1950s. An autopsy series of 82 vineyard workers exposed in Germany found 61 deaths from cancer, including 44 respiratory tract cancers; many skin cancers and Bowen disease also were reported.

The most common route of exposure to arsenic worldwide is through drinking water, with estimates of more than 50 million people using well water containing excess arsenic levels—20 million in Bangladesh alone. Chronic arsenic exposure has been linked to cancer of the bladder, lungs, skin, and kidney. Efforts to reduce exposure to arsenic-contaminated well water in Bangladesh have been attempted by the World Health Organization, the United Nations Educational, Scientific, and Cultural Organization (UNESCO), and other health organizations with limited success.

Nickel

The nickel refining industry was established in South Wales around 1900. During the subsequent 40 years, evidence accumulated for increased rates of nasal cancer, and later, lung cancer among workers in nickel refineries.⁷ Several prospective studies in England and Wales found elevated risks of nasal cancer (O/E = 12) and lung cancer (O/E = 16), particularly among process workers in the refinery. The highest cancer risk was associated with the calcination of impure nickel copper sulfate. Further studies indicated that the cancer risk began to decline when environmental controls were introduced in the industry during the 1930s. However, evidence of excess cancer risk associated with nickel exposure as late as the 1950s was reported in a refinery in Norway. Excessive rates of nasal, laryngeal, and lung cancers were observed; smoking and nickel exposure seemed to contribute to lung cancer risk in an additive manner.

Cadmium

An elevated risk of prostate and lung cancer among workers exposed to cadmium has been reported. Cadmium exposure has a stronger association with lung cancer than with prostate cancer.⁷ Several small historical prospective studies of cadmium smelters and battery workers show increased risks of prostate cancer (O/E = 1.2 to 3.5) and lung cancer (O/E = 1.35); however, a larger study of 7000 workers exposed to cadmium for at least 1 year showed no increased risk of prostate cancer. A major nonoccupational source of cadmium exposure is from cigarette smoke, which contains 1 to 2 µg of cadmium per pack of cigarettes.

Chromates

Several reports of lung cancer in chromate industry workers appeared in Germany during the 1930s. Subsequent epidemiological investigations examined this association in workers involved in chromate production, chromate pigment manufacture, and chrome plating.⁷ Excessive rates of lung cancer were found among workers in three chromate production plants in England (a threefold excess) and in a production plant in Baltimore (a twofold excess). Exposure to lead and zinc chromate pigments, but not to lead pigment alone, was related to excessive rates of lung cancer in a British study. Confirmatory results have been reported for workers exposed to lead and zinc chromates in the Netherlands, West Germany, and Norway. Experimental investigations indicate that the hexavalent salts of chromium are highly carcinogenic, whereas trivalent chromium is not carcinogenic.

Fibers and Dusts

Asbestos

The appearance of lung cancer in patients exposed to asbestosis was first reported in the 1930s. During the next 50 years, many different study complications were recognized and overcome in the process of determining the lung cancer risk associated with the naturally occurring silicate fiber, asbestos.⁵ A major problem was exposure assessment, in that many workers were mobile, having variable levels of exposure in a variety of industries or work sites. Furthermore, measurement techniques for asbestos fibers also were variable, making the use of historical measurements suspect. Some studies did not report asbestos type—that is, chrysotile, amosite, anthophyllite, or crocidolite. The long latency period between first exposure and lung cancer also complicates risk estimates. Misdiagnosis of the cancer pathologies specific to asbestos (mesothelioma) is a potential problem, especially in studies that depend on death certificates.

More than 30 epidemiological studies have been mounted to examine lung cancer risk in workers with potential exposure to asbestos during mining, milling, manufacturing, insulating, and shipbuilding.¹ In general, these studies demonstrate enhanced risk of lung cancer, and possibly enhanced risk of laryngeal cancer. The association between asbestos exposure and mesothelioma of the lung, a relatively rare cancer, was reported in the early 1960s. This finding was confirmed among insulation workers, asbestos manufacturing workers, and other populations exposed through their occupations.

Cases of pleural malignancies in Denmark and Germany in the 1930s were reported to be concentrated in seaport towns rather than in other urban areas. Later studies of the geographic distribution of mesothelioma deaths in England and Wales demonstrated a correlation between mesothelioma and areas of high asbestos use related to shipbuilding, gas mask manufacture, or asbestos manufacturing. Further studies led to the conclusion that the risk of mesothelioma was not limited to the workers but extended to members of their household and to residents living in the vicinity of asbestos-related industries.

Evidence of a synergistic effect of asbestos and smoking for lung cancer risk was found among insulation workers in New York.⁷ The age-standardized mortality ratio for lung cancer was 5.2 for nonsmoking asbestos-exposed workers (compared with nonexposed nonsmokers), 10.8 for nonexposed smokers, and 53.2 for asbestos-exposed smokers. Thus the risk of lung cancer from both smoking and exposure to asbestos is much greater than the sum of the risks associated with either exposure. A similar result was reported for female workers at a British asbestos factory, but the results for male workers at the same factory could not distinguish between additive or multiplicative effects.

Silica

Exposure to silica dusts occurs in several occupational groups, including foundry workers, pottery workers, miners, and quarry workers. Examination of occupational mortality statistics in high-exposure segments of these industries consistently show an increased risk for lung cancer among workers exposed to silica.¹ Two studies using silicosis registries have shown an association between silicosis and lung cancer. Excess mortality from lung cancer (O/E = 2.8) was found among 3600 men recorded in the Swedish silicosis registry from 1931 to 1969. An excess of lung cancer deaths (O/E = 2.0) also was found among 1910 miners registered with silicosis in Ontario, Canada, from 1940 to 1975. Pottery workers in the United States showed a significantly excessive rate of lung cancer among men whose work entailed making ceramic plumbing fixtures.

Wood Dust

The cancer risk associated with wood dust has been investigated in furniture workers, carpenters, woodworkers, lumberjacks, sawmill workers, and paper or pulp mill workers.⁷ Excessive rates of nasal adenocarcinoma are found consistently among furniture workers in several countries, primarily with exposure in the 1920s and 1930s. The highest risk is seen among those with exposure to hardwood dusts, and after a latent period of 30 years. Small increases in the risk of larynx cancer, lung cancer, and Hodgkin disease have also been reported among persons with these occupations.

Dietary Factors in Human Carcinogenesis: Naturally Occurring Carcinogens and Anticarcinogens

Most of the known human carcinogens discussed in the foregoing sections have been identified from occupational and iatrogenic exposures; however, with the exception of tobacco, such agents are rather minor contributors to the current overall cancer burden. Exposures to many of these agents typically have been at high doses in small, well-defined cohorts. Conversely, most human cancers probably result from interactions of several or more carcinogenic influences at low doses. The carcinogenic process is subject to influence by many modifying variables, which in the aggregate probably represent the largest determinant of human cancer. These variables can be constitutive, including age, gender, immunologic status, and genetic composition. Extraconstitutional variables also are important, particularly diet and lifestyle habits such as smoking and alcohol consumption. Many epidemiological studies indicate that general increases in consumption of fiber-rich cereals, fruits, and vegetables and decreased consumption of fat-rich foods and excessive alcohol will serve as prudent approaches to reducing overall cancer risk.

A variety of components in the diet can contribute to carcinogenesis. Two major influences are fat and calorie consumption. Fat consumption is most strongly associated with the hormone-dependent cancers (i.e., breast [Box 9-2], ovary, and endometrium in women and prostate in men) and gastrointestinal cancers (i.e., gallbladder, colon, and rectum in both sexes). It is not known to what degree these relationships are causal or if they relate to the type of fat (saturated, unsaturated, polyunsaturated) or the overall caloric content of the diet. Fats can promote tumor development directly and also are a major source of calories. Animals fed high-fat diets consistently demonstrate enhanced tumorigenic outcomes. Conversely, it has been recognized for decades that caloric restriction has a very powerful, general inhibitory effect on carcinogenesis in many induced and spontaneous laboratory animal tumor models. The behavioral changes required to effect comparable population-wide reductions in fat and/or calorie consumption, however, pose formidable challenges.

Box 9-2 Environmental Estrogens and Breast Cancer

Women with a high lifetime exposure to estrogen, such as that occurring with early onset of menarche and late menopause, may be at higher risk for breast cancer. In obese menopausal women, adipose tissue becomes the major source of estrogens, and higher blood estrogen levels among postmenopausal women have shown that higher levels are associated with increased risk of subsequent breast cancer.⁴² In this context, do xenoestrogens (synthetic chemicals that can act like estrogens) or other environmental factors such as phytoestrogens (plant estrogens) play a role in the risk for breast cancer? The possibility is certainly biologically plausible, because these molecules may interfere with the interactions of human estrogens with their receptors. In most cases these compounds are much weaker agonists; however, clear results are not in hand. For example, the Long Island Breast Cancer Study was unable to identify any environmental factors that could be responsible for the elevated incidence of breast cancer on Long Island or other locations.⁴³ No evidence emerged for organochlorine pesticides, which have long been known to have weak estrogenic activities. Bisphenol A, an industrial chemical present in some polycarbonate plastics used in bottles and infant food packaging, has come under recent scrutiny as a xenoestrogen.⁴⁴ Consumer education and industry action rather than regulatory mandates will reduce exposures to bisphenol A at this time. Some studies show that women who have diets high in phytoestrogens have lower rates of breast cancer. These phytoestrogens, unlike many xenoestrogens, are quickly broken down in the body. Thus they serve to oppose the actions of natural estrogen and sometimes lead to lower serum estrogen levels. The role of early life exposures as risk factors for late-in-life events is emerging as a key area of investigation in environmental carcinogenesis. Interplay between genetics, environmental factors, and age are all likely to create conditions favorable or unfavorable to cancer development.⁴²

The increasing prevalence of overweight and obese individuals worldwide, combined with mounting evidence for a clear association between obesity and cancer risk, presents a major public health concern. A recent analysis indicates that more than 900 million adults are overweight and almost 400 million are obese (BMI >30 kg/m²).¹⁶ In addition to the noncancer health risks associated with obesity, current evidence suggests that overall cancer mortality increases by 10% for each 5 kg/m² increase in BMI. The IARC concluded that the existing evidence base supports an association between obesity and cancers of the liver, colon, postmenopausal breast, endometrium, kidney, and esophagus.¹⁷ A recent meta-analysis of prospective studies found strong associations between increased BMI and esophageal adenocarcinoma, thyroid cancer, kidney cancer, and colon cancer among men and associations between increased BMI and endometrial cancer, gallbladder cancer, kidney cancer, and esophageal cancer among women.¹⁸

The mechanism by which obesity increases cancer risk is unclear at this time, although overfeeding in animal models clearly increases incidence of selected cancers. Possible general pathways include increased storage of lipophilic carcinogens in fatty tissue; enhanced metabolism and excretion of carcinogens due to exercise or increased physical activity; or altered gene expression related to physiological changes caused by obesity or activity level. However, it is likely that unique mechanisms are involved in each cancer site.¹⁹ For example, obesity leads to increased levels of circulating estradiol, which is a major risk factor for breast cancer. Obesity also increases gastroesophageal reflux, which is associated with esophageal cancer.

The risk of colon cancer is strongly associated with consumption of red meat and animal fat. This association has been observed in several international correlative studies and case-control studies. Prospective studies of colon cancer risk demonstrate an association with consumption of red meat and animal fat (but not with vegetable fat), independent of total energy intake.²⁰ The increased risk associated with animal fat primarily is due to meat intake as opposed to dairy product intake. Other studies that have addressed cooking practices have found that consumption of fried foods and barbecued, broiled, or smoked meats is associated with an increased risk of colorectal cancer. Several hypotheses have been proposed to explain the strong association of colon cancer risk with red meat and animal fat. Diets high in fat increase the incidence of chemically induced colon cancers in rats and also increase the excretion of primary bile acids, which are converted to secondary bile acids by bacterial metabolism. Some secondary bile acids (e.g., deoxycholic and lithocholic acid) are colon tumor promoters in animal models.

Thus it has been hypothesized that increased animal fat in the human diet leads to promotion of colon tumors by secondary bile acids. An alternative (or complementary) hypothesis is that carcinogens in cooked meats contribute to colon carcinogenesis. The cooking of meat produces at least two major classes of carcinogens—PAHs and heterocyclic aromatic amines—which induce genotoxic damage or cancer in the gastrointestinal tracts of animals receiving these compounds orally. Because cooked meats are a major source of animal fat in Western diets, it has been suggested that cooking-induced carcinogens in meat, rather than animal fat, play a causative role in colon cancer. Estimates of the average daily ingestion of PAHs and heterocyclic amines from diet are comparable (0.1 to 10 µg/day).

Highly mutagenic heterocyclic amine compounds are formed during the broiling or frying of meat and fish as a result of pyrolysis of amino acids and proteins. More than a dozen heterocyclic amines have been identified; the most common forms are quinolines, quinoxalines, pyridines, and carbolines.²¹ These compounds are carcinogenic in animals, causing colon cancer, mammary cancer, liver cancer, prostate cancer, and lymphoma. PhIP (2-amino-1-methyl-6-phenylimidazo(4,5-β)pyridine) is one of the most common heterocyclic amines formed in broiled and fried meats, occurring at levels comparable to or greater than those of benzo[a]pyrene. Colon adenocarcinomas develop in male rats that are fed PhIP, whereas mammary adenocarcinomas develop in female rats. Recent case-control studies of heterocyclic amine intake have found an increased risk for colon cancer among those who had the highest levels of heterocyclic amine intake.²²

Many minor dietary components act as carcinogens or anticarcinogens.²³ Dozens of natural mutagens and carcinogens derived from plant, fungal, and bacterial sources have been described, which present a significant carcinogenic challenge to humans. These natural carcinogens, however, are opposed by an equally expansive array of food-derived anticarcinogens. Such anticarcinogens consist of both nutrient (e.g., vitamins and minerals) and nonnutrient components, many of which function as antioxidants. In addition to scavenging oxidants, many of the nonnutrient anticarcinogens in plants alter the balance between the metabolic activation and inactivation of chemical carcinogens. Inverse epidemiological associations between risk of cancer at several sites and ingestion of fruits and vegetables have been observed.²⁴ These associations might be linked to β-carotene, other carotenoids, folate, fiber, vitamin C, and other antioxidants, and to other components of the fruits and vegetables. Although a protective role of specific nutrients has been difficult to establish in many instances, inverse relationships between intake of foods rich in vitamin C and the incidence of oral, esophageal, and gastric cancer have been described, as well as between β-carotene intake and lung cancer incidence. A fuller understanding of the role of dietary factors in human carcinogenesis is destined to have a significant impact on disease incidence.

Exposure Biomarkers and Susceptibility Factors

Assessing Human Exposure: Role for Intermediate Biomarkers

Increased understanding of the mechanistic basis of carcinogenesis provides opportunities for the identification of molecular biological markers that reflect events occurring between exposure and clinical disease. These molecular biological markers can be classified into three major categories:

1. Markers of exposure reflecting either an internal or a biologically effective dose of a carcinogen

2. Markers of effect indicating a biological response to an exposure

3. Markers of susceptibility that characterize the inherent susceptibility of an individual to a carcinogenic agent ²⁵

It is anticipated that the use of biological markers will help define the roles of environmental agents (particularly in complex mixtures) in the etiology of human cancer.

The interaction of a carcinogen with macromolecules was demonstrated by Miller and Miller in 1947, when they showed that azo dye bound to the liver protein of treated rats. Later studies indicated the importance of carcinogen modification of DNA (either directly or after metabolic activation) in the cancer process. The measurement of carcinogen metabolites, carcinogen-DNA adducts, or carcinogen-protein adducts in human tissues or fluids provide the basis for molecular dosimetry research and the rapidly expanding field of “molecular” cancer epidemiology. The potential advantage of this approach is that more accurate assessments of individual or group dose may be achieved than through estimates of carcinogen exposure in the environment or workplace.

Carcinogen-DNA and carcinogen-protein adducts have been detected in tissues from a variety of human populations with known or suspected carcinogen exposure.25,26 PAH-DNA adducts are elevated in white blood cells from persons occupationally exposed to airborne PAHs, including coke oven workers, foundry workers, and aluminum plant workers. Increased levels of PAH-DNA adducts also have been reported in lung tissue from heavy smokers. DNA isolated from exfoliated bladder epithelial cells of cigarette smokers has been shown to contain 4-aminobiphenyl adducts. Alkylation damage is detected in DNA from esophageal tissue of persons living in regions with documented dietary exposure to nitrosamines. Aflatoxin adducts are found in hepatic DNA after dietary exposure to this mycotoxin. Cisplatin-DNA intrastrand adducts have been measured in white blood cells of patients undergoing cancer chemotherapy.

Carcinogen-DNA adducts excreted in urine provide a noninvasive means for quantifying DNA damage. Urinary concentration of aflatoxin-guanine adducts is strongly correlated with dietary aflatoxin intake. In addition to their use as indicators of previous carcinogen exposure, DNA adducts have the potential for use as direct indicators of future cancer risk. The first prospective test of this application appeared in 1992, when detectable levels of aflatoxin-guanine adducts and several other aflatoxin metabolites in urine were shown to be predictive of the development of liver cancer.¹¹ Furthermore, an interactive effect for liver cancer risk was seen between urinary aflatoxin biomarkers and HBV infection history.

Carcinogen-protein adducts also have proved useful as biomarkers of human exposure. Alkylation damage in hemoglobin has been used as an exposure index in risk assessment analyses of persons occupationally exposed to ethylene oxide. It has been found that 4-aminobiphenyl adducts in hemoglobin are highly specific (and sensitive) markers of exposure to cigarette smoke.²⁵ The level of 4-aminobiphenyl hemoglobin adducts is related to the number of cigarettes smoked, the type of tobacco smoked, and the metabolic phenotype of the smoker. The levels of these adducts drop markedly after smoking cessation.

Unreacted carcinogen metabolites excreted in urine also are used to monitor human exposure to and uptake of carcinogens and related compounds. Concentrations of hydroxylated PAHs (e.g., 1-hydroxypyrene) in urine are elevated in smokers, in patients after topical treatment with coal tar, in road pavers, in coke oven workers, in aluminum plant workers, and in persons ingesting PAHs from food.²⁷ In occupational settings, the concentration of urinary 1-hydroxypyrene is highly correlated with estimated or measured concentrations of airborne PAHs. This noninvasive approach to exposure assessment has potential application as a routine biomonitoring tool. In all these studies, significant differences in adduct or metabolite levels often are observed between persons with similar carcinogen exposure. These differences have been attributed to several factors, including individual biological variability, exposure misclassification, and confounding variables such as diet, physical activity, smoking, or personal environment. An understanding of the true basis for these differences will be useful in elucidating the determinants of individual susceptibility to cancer. By identifying specific modulators that enhance or inhibit carcinogen metabolism and the formation of adducts, it will be possible to examine their effects on cancer risk in humans.

Genetic Polymorphisms and Human Susceptibility

Another major area of research in molecular cancer epidemiology is the study of individual metabolic phenotypes and their roles in determining biomarker levels and human susceptibility to carcinogens.²⁸ Many cytochrome P-450 metabolic enzymes are known to be involved in the activation of specific human carcinogens, and some have been linked to increased cancer risk. For example, inducible CYP1A1 activity is higher in cultured lymphocytes from persons with lung cancer than in control subjects. Genetic polymorphisms in the CYP1A1 structural gene also have been associated with lung cancer risk in Asian populations.

Hepatic arylamine N-acetyltransferase correlates with individual differences in susceptibility to bladder cancer in humans.²⁸ A series of genetic polymorphisms of this enzyme exist in humans, resulting in slow- and rapid-acetylator phenotypes. The rapid-acetylator phenotype seems to protect aromatic amine-exposed individuals from bladder cancer. These results are attributed to the competition of N-acetylation of arylamines against the formation of reactive arylamine metabolites that reach the bladder.

A meta-analysis of 22 case-control studies in the general population examined the effect of slow acetylation on risk of bladder cancer.⁹ Overall, the slow-acetylation phenotype/genotype was found to increase risk of bladder cancer by about 40% compared with rapid acetylators (odds ratio of 1.4, with a 95% confidence interval of 1.2 to 1.6). This report is the clearest example of a common metabolically modified cancer risk in the general population. The carcinogenic exposure in this case presumably is due to aryl amine compounds from various sources, including cigarette smoke.

An important mechanism that protects the genome from mutagenic effects of carcinogens is the repair of cellular DNA. The importance of this protection is perhaps best illustrated by the unusually severe effects of sunlight on persons who are deficient in DNA repair.²⁹ The rare inherited disorder xeroderma pigmentosum (XP) is characterized by various levels of DNA-repair deficiencies. Persons with XP show an unusually high incidence of multiple skin cancers and rarely live beyond early adulthood in the absence of protective measures against sunlight. The median age at onset for nonmelanoma skin cancers among persons with XP is approximately 8 years of age, compared with about 60 years of age in the general population. In addition, the prevalence of melanoma of the skin is unusually high among persons with XP.

Several other inherited diseases show altered cellular response to DNA damage.³⁰ Ataxia telangiectasia, Bloom syndrome, and Fanconi anemia are autosomal-recessive genetic disorders characterized by chromosomal instability, with malignancies developing in patients more frequently and at a younger age than occurs in the general population. Malignancies (primarily of the lymphoreticular system) develop in approximately 10% of persons with ataxia telangiectasia before the age of 20 years. Heterozygous relatives of persons with ataxia telangiectasia and those with Fanconi anemia also are at moderately increased risk of cancer compared with unrelated persons.

Although these diseases clearly represent unusual cases, some forms of these diseases (e.g., complementation group XP-E) show only moderate deficiencies in DNA repair—approximately 60% to 90% of normal—yet remain unusually susceptible to cancer. This finding suggests that small reductions in DNA repair efficiency can lead to considerable increases in cancer susceptibility. Interestingly, significant variability in DNA repair proficiency has been demonstrated among disease-free “normal” individuals. This variability in the general population could contribute to differences in susceptibility not only to skin cancer but also to cancers of other organs.

Public Health Approaches to Cancer Prevention

An aging population and a decline in mortality from cardiovascular disease could soon herald the emergence of cancer as the major cause of death in the United States. Moreover, the curative treatment of many established (and especially disseminated) malignancies remains an enigmatic problem for which progress is measured in small steps. Clearly, the optimal way for dealing with virtually all diseases, including cancer, is prevention. As a consequence, the challenge to public health professionals is to devise and implement preventive measures against cancer.

As presented in Figure 9-1, the prevention of cancer can take several forms. Primary prevention targets healthy persons and can be achieved by avoiding exposure to risk factors. Another approach is to stimulate the defense mechanisms of the host to interfere with the carcinogenic process. Secondary prevention measures use early detection and intervention before pathological conditions are clinically apparent. The success of these strategies will be greatly facilitated by the development of noninvasive biomarkers that identify high-risk individuals. Finally, tertiary prevention is aimed at minimizing the effect of existing disease and consequent disability by avoiding development of new cancers and complications or relapses after therapy for initial malignancies.

Figure 9-1 Strategies for prevention of multistage carcinogenesis.

Geographic Distribution of Cancer

The epidemiology of many cancer types is characterized by a striking heterogeneity of their geographic distribution. Significant differences in incidence occur, sometimes within small geographic areas, suggesting a strong role for external environmental or dietary causes. For example, cancer of the esophagus, the sixth most common cause of death from cancer worldwide,³¹ demonstrates a wide range of incidence. Most of these cancers (84%) occur in developing countries, where esophageal squamous cell carcinoma is the third most common cancer in men, after lung and stomach cancer. Populations with moderately elevated incidence rates of esophageal cancer (7 to 50/100,000/year) are found in South America, central Asia, Normandy France, and southern Africa and in African-American males, whereas extremely high incidence rates (50 to 200/100,000/year) are found among both men and women residing in parts of China³¹ and central Asia, stretching to the coast of the Caspian Sea.³² This incidence compares with low incidence areas (<5/100,000/year) such as western Africa and Central America. In some regions, putative causes for these “hot spots” have been identified: hot yerba maté consumption in Uruguay, Brazil, and northeast Argentina or tobacco and alcohol use in Normandy.³³ However, the causative factors in very high-risk areas, including China and Iran, remain unknown, and the identification of these agents is a high priority.

Other cancers demonstrating unique geographic distribution include biliary cancer (cholangiocarcinoma) associated with liver fluke infection in some Asian countries (especially Thailand) and African countries; gastric adenocarcinoma associated with virulent strains of Helicobacter pylori infection in South Korea, Japan, and parts of China; and hepatocellular carcinoma associated with HBV and aflatoxin exposure in parts of western Africa and eastern China.³¹ Clearly, preventive efforts will require individualized approaches given the variety of regional cancer challenges.