Tables

Cancer is a common disease in all countries of the world. The probability that a person will develop cancer by the age of 70 years, given survival to that age, varies between about 10 and 40% in both sexes. On average, in developed countries, about one person in five will die from cancer. This proportion is about one in 15 in developing countries. In this article, environmental cancer is defined as cancer caused (or prevented) by non-genetic factors, including human behaviour, habits, lifestyle and external factors over which the individual has no control. A stricter definition of environmental cancer is sometimes used, comprising only the effect of factors such as air and water pollution, and industrial waste.

Geographical Variation

Variation between geographical areas in the rates of particular types of cancer can be much greater than that for cancer as a whole. Known variation in the incidence of the more common cancers is summarized in table 1. The incidence of nasopharyngeal carcinoma, for example, varies some 500-fold between South East Asia and Europe. This wide variation in frequency of the various cancers has led to the view that much of human cancer is caused by factors in the environment. In particular, it has been argued that the lowest rate of a cancer observed in any population is indicative of the minimum, possibly spontaneous, rate occurring in the absence of causative factors. Thus the difference between the rate of a cancer in a given population and the minimum rate observed in any population is an estimate of the rate of the cancer in the first population which is attributable to environmental factors. On this basis it has been estimated, very approximately, that some 80 to 90% of all human cancers are environmentally determined (International Agency for Research on Cancer 1990).

Table 1.  Variation between populations covered by cancer registration in the incidence of common cancers.¹

¹ Data from cancer registries included in IARC 1992. Only cancer sites with cumulative rate larger or equal to 2% in the high-incidence area are included. Rates refer to males except for breast, cervix uteri, corpus uteri and ovary cancers.
² Cumulative rate % between 0 and 74 years of age.
Source: International Agency for Research on Cancer 1992.

There are, of course, other explanations for geographical variation in cancer rates. Under-registration of cancer in some populations may exaggerate the range of variation, but certainly cannot explain differences of the size shown in table 1. Genetic factors also may be important. It has been observed, however, that when populations migrate along a gradient of cancer incidence they often acquire a rate of cancer which is intermediate between that of their home country and that of the host country. This suggests that a change in environment, without genetic change, has changed the cancer incidence. For example, when Japanese migrate to the United States their rates of colon and breast cancer, which are low in Japan, rise, and their rate of stomach cancer, which is high in Japan, falls, both tending more closely towards United States’ rates. These changes may be delayed until the first post-migration generation but they still occur without genetic change. For some cancers, change with migration does not occur. For example, the Southern Chinese retain their high rate of cancer of the nasopharynx wherever they live, thus suggesting that genetic factors, or some cultural habit which changes little with migration, are responsible for this disease.

Time Trends

Further evidence of the role of environmental factors in cancer incidence has come from the observation of time trends. The most dramatic and well-known change has been the rise in lung cancer rates in males and females in parallel with but occurring some 20 to 30 years after the adoption of cigarette use, which has been seen in many regions of the world; more recently in a few countries, such as the United States, there has been the suggestion of a fall in rates among males following a reduction in tobacco smoking. Less well understood are the substantial falls in incidence of cancers including those of the stomach, oesophagus and cervix which have paralleled economic development in many countries. It would be difficult to explain these falls, however, except in terms of reduction in exposure to causal factors in the environment or, perhaps, increasing exposure to protective factors—again environmental.

Main Environmental Carcinogenic Agents

The importance of environmental factors as causes of human cancer has been further demonstrated by epidemiological studies relating particular agents to particular cancers. The main agents which have been identified are summarized in table 10. This table does not contain the drugs for which a causal link with human cancer has been established (such as diethylstilboestrol and several alkylating agents) or suspected (such as cyclophosphamide) (see also Table 9). In the case of these agents, the risk of cancer has to be balanced with the benefits of the treatment. Similarly, Table 10 does not contain agents that occur primarily in the occupational setting, such as chromium, nickel and aromatic amines. For a detailed discussion of these agents see the previous article “Occupational Carcinogens.” The relative importance of the agents listed in table 8 varies widely, depending on the potency of the agent and the number of people involved. The evidence of carcinogenicity of several environmental agents has been evaluated within the IARC Monographs programme (International Agency for Research on Cancer 1995) (see again “Occupational Carcinogens” for a discussion of the Monographs programme); table 10 is based mainly on the IARC Monograph evaluations. The most important agents among those listed in table 10 are those to which a substantial proportion of the population is exposed in relatively large amounts. They include particularly: ultraviolet (solar) radiation; tobacco smoking; alcohol drinking; betel quid chewing; hepatitis B; hepatitis C and human papilloma viruses; aflatoxins; possibly dietary fat, and dietary fiber and vitamin A and C deficiency; reproductive delay; and asbestos.

Attempts have been made to estimate numerically the relative contributions of these factors to the 80 or 90% of cancers which might be attributed to environmental factors. The pattern varies, of course, from population to population according to differences in exposures and possibly in the genetic susceptibility to various cancers. In many industrialized countries, however, tobacco smoking and dietary factors are likely to be responsible each for roughly one-third of environmentally determined cancers (Doll and Peto 1981); while in developing countries the role of biological agents is likely to be large and that of tobacco relatively small (but increasing, following the recent increase in the consumption of tobacco in these populations).

Interactions between Carcinogens

An additional aspect to consider is the presence of interactions between carcinogens. Thus for example, in the case of alcohol and tobacco, and cancer of the oesophagus, it has been shown that an increasing consumption of alcohol multiplies manyfold the rate of cancer produced by a given level of tobacco consumption. Alcohol by itself may facilitate transport of tobacco carcinogens, or others, into the cells of susceptible tissues. Multiplicative interaction may also be seen between initiating carcinogens, as between radon and its decay products and tobacco smoking in miners of uranium. Some environmental agents may act by promoting cancers which have been initiated by another agent—this is the most likely mechanism for an effect of dietary fat on the development of breast cancer (probably through increased production of the hormones which stimulate the breast). The reverse may also occur, as, for example, in the case of vitamin A, which probably has an anti-promoting effect on lung and possibly other cancers initiated by tobacco. Similar interactions may also occur between environmental and constitutional factors. In particular, genetic polymorphism to enzymes implicated in the metabolism of carcinogenic agents or DNA repair is probably an important requirement of individual susceptibility to the effect of environmental carcinogens.

The significance of interactions between carcinogens, from the point of view of cancer control, is that withdrawal of exposure to one of two (or more) interacting factors may give rise to a greater reduction in cancer incidence than would be predicted from consideration of the effect of the agent when acting alone. Thus, for example, withdrawal of cigarettes may eliminate almost entirely the excess rate of lung cancer in asbestos workers (although rates of mesothelioma would be unaffected).

Implications for Prevention

The realization that environmental factors are responsible for a large proportion of human cancers has laid the foundation for primary prevention of cancer by modification of exposure to the factors identified. Such modification may comprise: removal of a single major carcinogen; reduction, as discussed above, in exposure to one of several interacting carcinogens; increasing exposure to protective agents; or combinations of these approaches. While some of this may be achieved by community-wide regulation of the environment through, for example, environmental legislation, the apparent importance of lifestyle factors suggests that much of primary prevention will remain the responsibility of individuals. Governments, however, may still create a climate in which individuals find it easier to take the right decision.

Back

Occupational exposures account for only a minor proportion of the total number of cancers in the entire population. It has been estimated that 4% of all cancers can be attributed to occupational exposures, based on data from the United States, with a range of uncertainty from 2 to 8%. This implies that even total prevention of occupationally induced cancers would result in only a marginal reduction in national cancer rates.

However, for several reasons, this should not discourage efforts to prevent occupationally induced cancers. First, the estimate of 4% is an average figure for the entire population, including unexposed persons. Among people actually exposed to occupational carcinogens, the proportion of tumours attributable to occupation is much larger. Second, occupational exposures are avoidable hazards to which individuals are involuntarily exposed. An individual should not have to accept an increased risk of cancer in any occupation, especially if the cause is known. Third, occupationally induced cancers can be prevented by regulation, in contrast to cancers associated with lifestyle factors.

Prevention of occupationally induced cancer involves at least two stages: first, identification of a specific compound or occupational environment as carcinogenic; and second, imposing appropriate regulatory control. The principles and practice of regulatory control of known or suspected cancer hazards in the work environment vary considerably, not only among different parts of the developed and developing world, but also among countries of similar socio-economic development.

The International Agency for Research on Cancer (IARC) in Lyon, France, systematically compiles and evaluates epidemiological and experimental data on suspected or known carcinogens. The evaluations are presented in a series of monographs, which provide a basis for decisions on national regulations on the production and use of carcinogenic compounds (see “Occupational Carcinogens”, above.

Historical Background

The history of occupational cancer dates back to at least 1775, when Sir Percivall Pott published his classical report on scrotal cancer in chimney-sweeps, linking exposure to soot to the incidence of cancer. The finding had some immediate impact in that sweeps in some countries were granted the right to bathe at the end of the working day. Current studies of sweeps indicate that scrotal and skin cancer are now under control, although sweeps are still at increased risk for several other cancers.

In the 1890s, a cluster of bladder cancer was reported at a German dye factory by a surgeon at a nearby hospital. The causative compounds were later identified as aromatic amines, and these now appear in lists of carcinogenic substances in most countries. Later examples include skin cancer in radium-dial painters, nose and sinus cancer among woodworkers caused by inhalation of wood dust, and “mule-spinner’s disease”—that is, scrotal cancer among cotton industry workers caused by mineral oil mist. Leukaemia induced by exposure to benzene in the shoe repair and manufacturing industry also represents a hazard that has been reduced after the identification of carcinogens in the workplace.

In the case of linking asbestos exposure to cancer, this history illustrates a situation with a considerable time-lag between risk identification and regulatory action. Epidemiological results indicating that exposure to asbestos was associated with an increased risk of lung cancer were already starting to accumulate by the 1930s. More convincing evidence appeared around 1955, but it was not until the mid-1970s that effective steps for regulatory action began.

The identification of the hazards associated with vinyl chloride represents a different history, where prompt regulatory action followed identification of the carcinogen. In the 1960s, most countries had adopted an exposure limit value for vinyl chloride of 500 parts per million (ppm). In 1974, the first reports of an increased frequency of the rare tumour liver angiosarcoma among vinyl chloride workers were soon followed by positive animal experimental studies. After vinyl chloride was identified as carcinogenic, regulatory actions were taken for a prompt reduction of the exposure to the current limit of 1 to 5 ppm.

Methods Used for the Identificationof Occupational Carcinogens

The methods in the historical examples cited above range from observations of clusters of disease by astute clinicians to more formal epidemiological studies—that is, investigations of the disease rate (cancer rate) among human beings. Results from epidemiological studies are of high relevance for evaluations of the risk to humans. A major drawback of cancer epidemiological studies is that a long time period, usually at least 15 years, is necessary to demonstrate and evaluate the effects of an exposure to a potential carcinogen. This is unsatisfactory for surveillance purposes, and other methods must be applied for a quicker evaluation of recently introduced substances. Since the beginning of this century, animal carcinogenicity studies have been used for this purpose. However, the extrapolation from animals to humans introduces considerable uncertainty. The methods also have limitations in that a large number of animals must be followed for several years.

The need for methods with a more rapid response was partly met in 1971, when the short-term mutagenicity test (Ames test) was introduced. This test uses bacteria to measure the mutagenic activity of a substance (its ability to cause irreparable changes in the cellular genetic material, DNA). A problem in the interpretation of the results of bacterial tests is that not all substances causing human cancers are mutagenic, and not all bacterial mutagens are considered to be cancer hazards for human beings. However, the finding that a substance is mutagenic is usually taken as an indication that the substance might represent a cancer hazard for humans.

New genetic and molecular biology methods have been developed during the last 15 years, with the aim of detecting human cancer hazards. This discipline is termed “molecular epidemiology.” Genetic and molecular events are studied in order to clarify the process of cancer formation and thus develop methods for early detection of cancer, or indications of increased risk of the development of cancer. These methods include analysis of damage to the genetic material and the formation of chemical linkages (adducts) between pollutants and the genetic material. The presence of chromosomal aberrations clearly indicates effects on the genetic material which may be associated with cancer development. However, the role of molecular epidemiological findings in human cancer risk assessment remains to be settled, and research is under way to indicate more clearly exactly how results of these analyses should be interpreted.

Surveillance and Screening

The strategies for prevention of occupationally induced cancers differ from those applied for control of cancer associated with lifestyle or other environmental exposures. In the occupational field, the main strategy for cancer control has been reduction or total elimination of exposure to cancer-causing agents. Methods based on early detection by screening programmes, such as those applied for cervical cancer or breast cancer, have been of very limited importance in occupational health.

Surveillance

Information from population records on cancer rates and occupation may be used for surveillance of cancer frequencies in various occupations. Several methods to obtain such information have been applied, depending on the registries available. The limitations and possibilities depend largely on the quality of the information in the registries. Information on disease rate (cancer frequency) is typically obtained from local or national cancer registries (see below), or from death certificate data, while information on the age-composition and size of occupational groups is obtained from population registries.

The classical example of this type of information is the “Decennial supplements on occupational mortality,” published in the UK since the end of the nineteenth century. These publications use death certificate information on cause of death and on occupation, together with census data on frequencies of occupations in the entire population, to calculate cause-specific death rates in different occupations. This type of statistic is a useful tool to monitor the cancer frequency in occupations with known risks, but its ability to detect previously unknown risks is limited. This type of approach may also suffer from problems associated with systematic differences in the coding of occupations on the death certificates and in the census data.

The use of personal identification numbers in the Nordic countries has offered a special opportunity to link individual census data on occupations with cancer registration data, and to directly calculate cancer rates in different occupations. In Sweden, a permanent linkage of the censuses of 1960 and 1970 and the cancer incidence during subsequent years have been made available for researchers and have been used for a large number of studies. This Swedish Cancer-Environment Registry has been used for a general survey of certain cancers tabulated by occupation. The survey was initiated by a governmental committee investigating hazards in the work environment. Similar linkages have been performed in the other Nordic countries.

Generally, statistics based on routinely collected cancer incidence and census data have the advantage of ease in providing large amounts of information. The method gives information on the cancer frequencies regarding occupation only, not in relation to certain exposures. This introduces a considerable dilution of the associations, since exposure may differ considerably among individuals in the same occupation. Epidemiological studies of the cohort type (where the cancer experience among a group of exposed workers is compared with that in unexposed workers matched for age, sex and other factors) or the case-control type (where the exposure experience of a group of persons with cancer is compared to that in a sample of the general population) give better opportunities for detailed exposure description, and thus better opportunities for investigation of the consistency of any observed risk increase, for example by examining the data for any exposure-response trends.

The possibility of obtaining more refined exposure data together with routinely collected cancer notifications was investigated in a prospective Canadian case-control study. The study was set up in the Montreal metropolitan area in 1979. Occupational histories were obtained from males as they were added to the local cancer registry, and the histories were subsequently coded for exposure to a number of chemicals by occupational hygienists. Later, the cancer risks in relation to a number of substances were calculated and published (Siemiatycki 1991).

In conclusion, the continuous production of surveillance data based on recorded information provides an effective and comparatively easy way to monitor cancer frequency by occupation. While the main purpose achieved is surveillance of known risk factors, the possibilities for the identification of new risks are limited. Registry-based studies should not be used for conclusions regarding the absence of risk in an occupation unless the proportion of individuals significantly exposed is more precisely known. It is quite common that only a relatively small percentage of members of an occupation actually are exposed; for these individuals the substance may represent a substantial hazard, but this will not be observable (i.e., will be statistically diluted) when the entire occupational group is analysed as a single group.

Screening

Screening for occupational cancer in exposed populations for purposes of early diagnosis is rarely applied, but has been tested in some settings where exposure has been difficult to eliminate. For example, much interest has focused on methods for early detection of lung cancer among people exposed to asbestos. With asbestos exposures, an increased risk persists for a long time, even after cessation of exposure. Thus, continuous evaluation of the health status of exposed individuals is justified. Chest x rays and cytological investigation of sputum have been used. Unfortunately, when tested under comparable conditions neither of these methods reduces the mortality significantly, even if some cases may be detected earlier. One of the reasons for this negative result is that the prognosis of lung cancer is little affected by early diagnosis. Another problem is that the x rays themselves represent a cancer hazard which, while small for the individual, may be significant when applied to a large number of individuals (i.e., all those screened).

Screening also has been proposed for bladder cancer in certain occupations, such as the rubber industry. Investigations of cellular changes in, or mutagenicity of, workers’ urine have been reported. However, the value of following cytological changes for population screening has been questioned, and the value of the mutagenicity tests awaits further scientific evaluation, since the prognostic value of having increased mutagenic activity in the urine is not known.

Judgements on the value of screening also depend on the intensity of the exposure, and thus the size of the expected cancer risk. Screening might be more justified in small groups exposed to high levels of carcinogens than among large groups exposed to low levels.

To summarize, no routine screening methods for occupational cancers can be recommended on the basis of present knowledge. The development of new molecular epidemiological techniques may improve the prospects for early cancer detection, but more information is needed before conclusions can be drawn.

Cancer Registration

During this century, cancer registries have been set up at several locations throughout the world. The International Agency for Research on Cancer (IARC) (1992) has compiled data on cancer incidence in different parts of the world in a series of publications, “Cancer Incidence in Five Continents.” Volume 6 of this publication lists 131 cancer registries in 48 countries.

Two main features determine the potential usefulness of a cancer registry: a well-defined catchment area (defining the geographical area involved), and the quality and completeness of the recorded information. Many of those registries that were set up early do not cover a geographically well-defined area, but rather are confined to the catchment area of a hospital.

There are several potential uses of cancer registries in the prevention of occupational cancer. A complete registry with nationwide coverage and a high quality of recorded information can result in excellent opportunities for monitoring the cancer incidence in the population. This requires access to population data to calculate age-standardized cancer rates. Some registries also contain data on occupation, which therefore facilitates the monitoring of cancer risk in different occupations.

Registries also may serve as a source for the identification of cases for epidemiological studies of both the cohort and case-control types. In the cohort study, personal identification data of the cohort is matched to the registry to obtain information on the type of cancer (i.e., as in record linkage studies). This assumes that a reliable identifying system exists (for example, personal identification numbers in the Nordic countries) and that confidentiality laws do not prohibit use of the registry in this way. For case-control studies, the registry may be used as a source for cases, although some practical problems arise. First, the cancer registries cannot, for methodological reasons, be quite up to date regarding recently diagnosed cases. The reporting system, and necessary checks and corrections of the obtained information, results in some lag time. For concurrent or prospective case-control studies, where it is desirable to contact the individuals themselves soon after a cancer diagnosis, it usually is necessary to set up an alternative way of identifying cases, for example via hospital records. Second, in some countries, confidentiality laws prohibit the identification of potential study participants who are to be contacted personally.

Registries also provide an excellent source for calculating background cancer rates to use for comparison of the cancer frequency in cohort studies of certain occupations or industries.

In studying cancer, cancer registries have several advantages over mortality registries commonly found in many countries. The accuracy of the cancer diagnoses is often better in cancer registries than in mortality registries, which are usually based on death certificate data. Another advantage is that the cancer registry often holds information on histological tumour type, and also permits the study of living persons with cancer, and is not limited to deceased persons. Above all, registries hold cancer morbidity data, permitting the study of cancers that are not rapidly fatal and/or not fatal at all.

Environmental Control

There are three main strategies for reducing workplace exposures to known or suspected carcinogens: elimination of the substance, reduced exposure by reduced emission or improved ventilation, and personal protection of the workers.

It has long been debated whether a true threshold for carcinogen exposure exists, below which no risk is present. It is often assumed that the risk should be extrapolated linearly down to zero risk at zero exposure. If this is the case, then no exposure limit, no matter how low, would be considered entirely risk-free. Despite this, many countries have defined exposure limits for some carcinogenic substances, while, for others, no exposure limit value has been assigned.

Elimination of a compound may give rise to problems when replacement substances are introduced and when the toxicity of the replacement substance must be lower than that of the substance replaced.

Reducing the exposure at the source may be relatively easily accomplished for process chemicals by encapsulation of the process and ventilation. For example, when the carcinogenic properties of vinyl chloride were discovered, the exposure limit value for vinyl chloride was lowered by a factor of one hundred or more in several countries. Although this standard was at first considered impossible to achieve by industry, later techniques allowed compliance with the new limit. Reduction of exposure at the source may be difficult to apply to substances that are used under less controlled conditions, or are formed during the work operation (e.g., motor exhausts). The compliance with exposure limits requires regular monitoring of workroom air levels.

When exposure cannot be controlled either by elimination or by reduced emissions, the use of personal protection devices is the only remaining way to minimize the exposure. These devices range from filter masks to air-supplied helmets and protective clothing. The main route of exposure must be considered in deciding appropriate protection. However, many personal protection devices cause discomfort to the user, and filter masks introduce an increased respiratory resistance which may be very significant in physically demanding jobs. The protective effect of respirators is generally unpredictable and depends on several factors, including how well the mask is fitted to the face and how often filters are changed. Personal protection must be considered as a last resort, to be attempted only when more effective ways of reducing exposure fail.

Research Approaches

It is striking how little research has been done to evaluate the impact of programmes or strategies to reduce the risk to workers of known occupational cancer hazards. With the possible exception of asbestos, few such evaluations have been conducted. Developing better methods for control of occupational cancer should include an evaluation of how present knowledge is actually put to use.

Improved control of occupational carcinogens in the workplace requires the development of a number of different areas of occupational safety and health. The process of identification of risks is a basic prerequisite for reducing exposure to carcinogens in the workplace. Risk identification in the future must solve certain methodological problems. More refined epidemiological methods are required if smaller risks are to be detected. More precise data on exposure for both the substance under study and possible confounding exposures will be necessary. More refined methods for description of the exact dose of the carcinogen delivered to the specific target organ also will increase the power of exposure-response calculations. Today, it is not uncommon that very crude substitutes are used for the actual measurement of target organ dose, such as the number of years employed in the industry. It is quite clear that such estimates of dose are considerably misclassified when used as a surrogate for dose. The presence of an exposure-response relationship is usually taken as strong evidence of an aetiological relationship. However, the reverse, lack of demonstration of an exposure-response relationship, is not necessarily evidence that no risk is involved, especially when crude measures of target organ dose are used. If target organ dose could be determined, then actual dose-response trends would carry even more weight as evidence for causation.

Molecular epidemiology is a rapidly growing area of research. Further insight into the mechanisms of cancer development can be expected, and the possibility of the early detection of carcinogenic effects will lead to earlier treatment. In addition, indicators of carcinogenic exposure will lead to improved identification of new risks.

Development of methods for supervision and regulatory control of the work environment are as necessary as methods for the identification of risks. Methods for regulatory control differ considerably even among western countries. The systems for regulation used in each country depend largely on socio-political factors and the status of labour rights. The regulation of toxic exposures is obviously a political decision. However, objective research into the effects of different types of regulatory systems could serve as a guide for politicians and decision-makers.

A number of specific research questions also need to be addressed. Methods to describe the expected effect of withdrawal of a carcinogenic substance or reduction of exposure to the substance need to be developed (i.e., the impact of interventions must be assessed). The calculation of the preventive effect of risk reduction raises certain problems when interacting substances are studied (e.g., asbestos and tobacco smoke). The preventive effect of removing one of two interacting substances is comparatively greater than when the two have only a simple additive effect.

The implications of the multistage theory of carcinogenesis for the expected effect of withdrawal of a carcinogen also adds a further complication. This theory states that the development of cancer is a process involving several cellular events (stages). Carcinogenic substances may act either in early or late stages, or both. For example, ionizing radiation is believed to affect mainly early stages in inducing certain cancer types, while arsenic acts mainly at late stages in lung cancer development. Tobacco smoke affects both early and late stages in the carcinogenic process. The effect of withdrawing a substance involved in an early stage would not be reflected in a reduced cancer rate in the population for a long time, while the removal of a “late-acting” carcinogen would be reflected in a reduced cancer rate within a few years. This is an important consideration when evaluating the effects of risk-reduction intervention programmes.

Finally, the effects of new preventive factors have recently attracted considerable interest. During the last five years, a large number of reports have been published on the preventive effect on lung cancer of consuming fruits and vegetables. The effect seems to be very consistent and strong. For example, the risk of lung cancer has been reported as double among those with a low consumption of fruits and vegetables versus those with high intake. Thus, future studies of occupational lung cancer would have greater precision and validity if individual data on fruit and vegetable consumption can be included in the analysis.

In conclusion, improved prevention of occupational cancer involves both improved methods for risk identification and more research on the effects of regulatory control. For risk identification, developments in epidemiology should mainly be directed toward better exposure information, while in the experimental field, validation of the results of molecular epidemiological methods regarding cancer risk are needed.

Back