Global cancer burden

The existence of differing patterns of cancer occurrence throughout the world has been known for many years, as perusal of the fascinating monograph published in 1915 by Hoffman, the actuary to the Prudential Assurance Company reveals. It was not, however, until 1984 that estimates (based on the year 1975) were made, not only of the global cancer burden but also for each of the 24 demographic regions recognized by the United Nations (UNO, 1986). This was later updated by Parkin et al. (1988), who estimated the global total of new cancers to be 6.35 million. This number was almost exactly divided between the sexes and between developed and developing countries, although two-thirds of the world population dwells in the latter.

The ‘top ten’ cancers are given in Table 22.1. Unfortunately, these figures are essentially estimates, since they are not uniformly collected, being derived from various sources. The latter include incidence figures from cancer registries, and incidence estimates extrapolated from mortality. For large areas of the world, numbers are based on relative frequency data, mostly obtained from the monograph Cancer Occurrence in Developing Countries (Parkin, 1986). Regrettably, the calculation of age-specific and age-standardized incidence rates, which provide a measure of the difference in risk between populations was not possible. The methods used, notably for the developing countries, are discussed in the original papers.

Introduction

It has been known for many years that two or more independent primary tumors can arise in the same patient. Schoenberg (1977), who surveyed multiple cancers in Connecticut between 1935–1964, appears to have been the first to make a comprehensive quantitative analysis of the risk of developing a second tumor through the investigation of 120, 195 cancer patients from a well-defined population.

Second cancers may arise in the same individual for a number of reasons (Table 59.1). The major issue is usually whether an excess risk represents a common causal factor or some degree of genetic susceptibility. Among relevant issues are the following: (i) the recognition of syndromes of associated sites which may suggest a common etiology; (ii) identification of the regions of the body to which special attention should be paid when following up individuals with the first cancer; (iii) assessment of the carcinogenicity of treatments used for a first cancer.

Common etiological factors

The combined effects of tobacco and alcohol largely account for the constellation of cancers arising in the oral cavity, larynx and oesophagus. In contrast the association between retinoblastoma and osteosarcoma is most likely to reflect the influence of common hereditary factors. The interplay of both exogenous and endogenous factors in multiple cancers of the breast, uterine corpus, ovary and colon has long intrigued investigators. Both reproductive factors and dietary habits are believed to be involved. Biilow et al. (1990) report metachronous colorectal cancers can be as high as 30% after 40 years.

Introduction

While cancers of the colon and rectum show some differences in their geographic and temporal distribution, it is usual to discuss them together in terms of causation.

Histology, classification and diagnosis

Epithelial tumors of the colorectum are adenocarcinomas. In many patients, such cancers are frequently associated with benign polyps and adenomas as well as dysplastic mucosal lesions which are believed to be precancerous (Ponz de Leon et al., 1987).

Geographic differences have been observed in the distribution by subsite (Schottenfeld & Winawer, 1982). Unfortunately, the anatomical boundary around the recto-sigmoid (which for international classification purposes is assigned to the rectum), is variable and the terms sigmoid and recto-sigmoid are often used interchangeably by clinicians. De Jong et al., (1972) concluded, however, that this made little difference to incidence rates. Non-epithelial tumors are discussed in Chapter 37.

Descriptive epidemiology

Incidence

For colon cancer, the highest male incidence rates are in the white population of Connecticut, USA (34.1). The highest female rate (29.0) is in the black population of Detroit (Fig. 28.1). Rates, in the range of 15 to 20, are observed in Canada and much of western Europe. In western Europe, the incidence tends to be more than twice that east of the Elbe River. In Asia and Latin America, rates are low to intermediate (10 to 15).

Introduction

Brain (ICD-9 191) and nervous system (ICD-9 192), cover a range of neoplasms arising from glial tissue and the meninges.

Histology, classification and diagnosis

Histological classification is based on the presumptive cell of origin (Rubinstein, 1972), but there are often diagnostic difficulties in determining type. Glial tumors show wide variations in aggressiveness, some being highly malignant whereas others grow relatively slowly. There are also differences in the age distribution of the various histological types, medulloblastoma being found exclusively in children (Table 51.1 and 51.2). Signs and symptoms depend on the size, rate of growth and anatomical location of the tumors. Clinically, tumors of the meninges and brain may present with similar findings, so that earlier statistics on different types may be biased. While glial tumors never metastasize, many other tumors frequently metastasize to the brain, notably bronchogenic carcinoma, and breast cancer in females. Metastases may also be present in the bones of the skull and vertebral column and thus indirectly mimic CNS lesions.

The availability of medical care clearly influences the likelihood of diagnosis, and in the past intracranial tumors were underdiagnosed. A full autopsy is often necessary for accurate diagnosis. Given the difficulty of obtaining biopsy material there may be an element of uncertainty about the diagnosis in several registries. Newer imaging procedures should increase the accuracy of registration statistics. Helseth et al. (1988) provide an excellent discussion of these problems.

Introduction

Globally, breast cancer is the most frequent malignancy among women with an estimated 572,000 new cases in 1980 or 18.4% of all cancers. The disease is extremely rare in males: in females it shows marked geographical variation.

Histology, classification and diagnosis

Most breast tumors arise from the ductal or lobular epithelial cells. A number of histological subgroups are described which, in general, do not appear to have etiological relevance although differences in biological behavior between Japanese and North American women have been reported. Breast biopsies have increased with growth of mammography with a resultant rise in the diagnosis of borderline in situ lesions which complicates interpretation of incidence and mortality trends. Mortality is also influenced by newer therapeutic techniques. The risk of cancer developing in patients with benign breast disease is directly related to the degree of epithelial atypia. While hormonal receptors appear significant for prognosis, they appear to be unrelated to specific risk factors.

Descriptive epidemiology

Incidence and mortality

The highest rates occur in North America, Israel and temperate South America (over 60). High rates (40-60) are also seen in north western Europe and in Australia. Intermediate rates (20–40) are found in parts of Eastern Europe, tropical South America, Philippines and Singapore. Low rates (less than 20) are seen in China, parts of Japan and India and in Kuwaitis. In the USA rates are higher in whites than in blacks.

Introduction

This glossary contains a sample of the terms used in epidemiology. The definitions given below are based on those provided by Last (1988). For further details, other sources should be consulted (Stedman, 1982; Friedman, 1980; Lilienfeld & Lilienfeld, 1980; Kendall & Buckland, 1982; Abramson, 1984; Murphy, 1985; Pressat, 1985; Last, 1986; Schuman, 1986; Jammal et al., 1988).

Glossary of terms

Adjustment a procedure in which potentially misleading effects of differences in the composition of populations being compared have been minimized – as much as is possible – by statistical methods. The mathematical procedure commonly used to adjust incidence or mortality rates for age differences is direct or indirect standardization.

Age-specific rate a rate calculated for a specific age-group, i.e., the numerator and denominator refer to the same age group. A multiplier, usually 100,000, is chosen to produce a rate that can be expressed as a convenient number.

Analytical study a study designed to examine hypothesized causal relationships. An analytical study is usually concerned with identifying the health effects of specific risk factors. In this type of study, individuals are classified either according to the absence or presence of a specific disease and/or according to the presence or absence of risk factors presumed to influence disease occurrence.

Introduction

Adenocarcinoma of the pancreas is the fifteenth most common cancer in the world, and is more frequent in developed, than in developing, countries. Tobacco is the most important etiological factor so far identified.

Histology, classification and diagnosis

Tumors of the pancreas can be divided into common adenocarcinomas and infrequent islet cell tumors of the endocrine pancreas. Many pancreatic tumors were formerly attributed to the gastro-intestinal tract or the liver, which may partly explain the reported incidence rise since 1950.

Descriptive epidemiology

Incidence

The incidence and distribution of adenocarcinomas has been reviewed by Mack (1982). Geographic variations are not as great as for other gastrointestinal cancers. Among males, high rates (over 10) are seen in American blacks, Maoris in New Zealand, and in some European countries; intermediate rates (over 4) in whites in North America, the rest of Europe, Australia, Japan, China, Israel and in most South American countries; and low rates (less than 4) in Africa, India and some South American countries (Figure 31.1). (The rate in male Koreans in Los Angeles is based on 11 cases.) In the USA, rates are higher in blacks than in whites, but are relatively low in Seventh Day Adventists and Mormons. A consistent association with social class has not been observed. Higher frequencies in urban than in rural areas may be a diagnostic artefact.

Introduction

Most known human chemical carcinogens were first identified in the work-place generally due to astute clinical observation following the tradition of Percival Pott (Historical introduction). By the early 1970s, most of the major occupational carcinogenic hazards known today had been recognized (McClure & MacMahon, 1980). Since then, although many industrial processes or compounds have been suspected to carry an elevated cancer risk and extensively studied, rarely has the evidence for humans been sufficient to implicate a previously unknown agent. The burden of occupational cancer in industrial states is generally considered as less than 6% (Wynder & Gori, 1977; Higginson & Muir, 1979; Doll & Peto, 1981) (Conclusions). Smaller estimates have been made in Finland (0.5%) (Hemminki & Vainio, 1984), Japan and other countries.

The period since World War II has seen major improvements in industrial hygienic practices in many countries, so that certain hazardous exposures should have been reduced. Thus, risk estimates made in the late 1970s largely reflect exposures prevalent in earlier decades and may be out of date. None the less, the problem of occupational cancer remains important especially in developing countries which offer a unique opportunity for control and prevention. Methodology and limitations in measuring the burden of occupational cancer are discussed in Banbury Report 9 (Peto & Schneiderman, 1981).

Scars and foreign bodies

Epithelial tumors and fibrous scarring

Carcinomas arise in relation to scar formation as observed in schistosomiasis and asbestosis, but it is not clear to what extent the scar formation per se actually contributes to these tumors. Several reports suggest that squamous cell carcinoma of the skin may arise as a complication of chronic ulcers, chronic inflammatory skin diseases, burns and scars or even a single injury. Traditionally this complication has been reported mainly in dark-skinned populations from Africa and Asia (Camain et al., 1972; Mulay, 1963), but has also been described in black Americans (Fleming et al., 1975; Keith et al., 1980). Case reports suggest that skin cancer occasionally complicates the scars of chronic skin diseases such as tuberculosis, leprosy, syphilis and discoid lupus (Keith et al., 1980). Both basocellular carcinoma and dermatofibroma have been reported to develop in a smallpox vaccination scar (Ribeiro et al., 1988).

In India, kangri cancer occurs in burn scars in the lower abdomen and thighs of people who use baskets (kangri) containing clay pots with burning charcoal to warm their bodies in winter (Mulay, 1963).

Soft tissue sarcomas (STS)

Foreign bodies

In rodents, the experimental transplant of a membrane, irrespective of its chemical nature, e.g. glass coverslip or plastic sheet, leads frequently to capsule formation followed by a soft tissue sarcoma.

Introduction

The non-melanoma skin (NMS) cancers, principally basal cell and squamous cell carcinomas, are very common but rarely fatal.

Histology, classification and diagnosis

Basal cell carcinomas are usually found on the face above a line drawn from the angle of the mouth to the ear lobe. Although slow growing, and almost never mestastasizing, these cancers, if neglected, can result in large ulcers hence the term ‘rodent ulcer’. Squamous cell carcinomas can occur on any part of the body, especially in the dorsum of the hand, or on the legs (notably in Africa). Other skin cancers, mainly arising from sweat glands, are very rare.

Descriptive epidemiology

Incidence

The incidence of NMS tumors shows wide variation. The true incidence is difficult to determine in view of their high frequency, low lethality and the fact that an individual may have several over a lifetime. Many cancer registries do not record basal cell carcinomas, which are also often under reported.

The highest rates of NMS cancer recently reported are from Tasmania in Australia (167.2 in males and 89.3 in females) and from British Columbia in Canada (109.1 in males and 75.5 in females). In El Paso in Texas, after a special effort to assess the true incidence, rates were as high as 144.9 in ‘Anglo’ males and 73.3 in ‘Anglo’ females (Waterhouse et al., 1976). Rates in the 50 to 60 range are observed in males in Brazil, Colombia and Switzerland.

This monograph attempts to summarize the epidemiology of human cancer, including not only its geographical distribution in various environments and ethnic groups, but also known or suspected causes. The aim is to make available a compact source of data and a summary consensus of current views on etiology and thus provide a background, not only for evaluating cancer control and preventive policies, but also for developing strategies for research into the causes of specific tumors. The authors wish to emphasize that epidemiology remains the only discipline which can provide definitive data as to the existence, extent and nature of cancer risks in humans and the impact of preventive and intervention strategies. The term environmental carcinogenesis, as used here, reflects the Hippocratic view that the environment covers all exogenous factors that impinge on human health, whether they be physical, biological, cultural or chemical. The techniques used by epidemiologists to identify and measure the determinants of human neoplastic disease are described briefly.

No attempt is made to explore in depth the sophisticated biomathematical techniques now used in the analysis of epidemiological data. Their correct application is not simple and requires collaboration between the epidemiologist and biostatistician. The mere statement that an observation is statistically significant has little meaning unless the many variables that may complicate a study have been taken into account. These issues have been extensively reviewed by Breslow and Day (1980, 1987).

Introduction

Certain types of ultraviolet (UV) light are strongly carcinogenic. Yet complete avoidance of exposure is almost impossible. Further, ultraviolet light is indispensable, permitting the skin to synthesize vitamin D, complementing that derived from dietary sources.

Spectrum of ultraviolet light

The UV portion of the electromagnetic spectrum occupies the wavelengths between 200 nm and 400 nm. It is customary to distinguish between UV-C (200–280 nm), UV-B (280–320 nm) and UV-A (320–400 nm), each having different effects on the skin and eye. The shorter the wavelength, the more destructive the radiation.

Ultraviolet light below 295 nm is totally absorbed by the oxygen and ozone of the atmosphere, that between 295 and 329 nm being partially absorbed. The more oblique the sun's rays, the greater thickness of atmosphere which has to be traversed and hence the shorter rays are diffused or absorbed. Considerable amounts of ultraviolet light may be reflected, thus the combination of snow and blue sky may expose the skin to as much UV-B (310 nm) as comes from the sky directly.

Effects

The amount of UV light reaching the melanocyte varies by wavelength. UV-A penetrates into the dermis, between 1 to 10% reaching the basal layer, a depth practically never reached by UV-C. Heavy skin pigmentation is, of course, protective. Sunburn is maximal about 24 hours after exposure to UV-B, occurring more rapidly than for UV-A.