Martha S. Linet, M.D., M.P.H.*
From the Biostatistics Branch, Division of Cancer Etiology,
National Cancer Institute, Bethesda, Maryland
The leukemias are a family of biologically diverse malignancies resulting from an abnormal change in an early form of one or a few blood cells that arise in the bone marrow. The abnormal leukemic cells progressively increase in number, and suppress growth of normal blood cells (Henderson, 1990). Clinical symptoms that result from loss of normal blood cell function include infections, fever, abnormal bruising or bleeding, and fatigue. Other symptoms are directly related to the progressive increase in the number of leukemic cells in the spleen, liver, or lymph nodes.
Five main types (and an increasing number of subtypes) of leukemia have been identified:
1) acute lymphocytic leukemia (ALL);
2) chronic myelocytic or granulocytic leukemia (CML or CGL);
3) acute myelocytic or acute nonlymphocytic leukemia (AML or ANLL);
4) chronic lymphocytic leukemia (CLL), and,
5) adult T-cell leukemia (ATL) (Linet, 1985).
Together these account for about 2.5 percent of the total annual cancer incidence in the United States and about one-third of cancers in children (Miller et al., 1993).
There are notable differences in age distribution by subtype. Acute lymphocytic, the most common childhood cancer in most Western countries, is low in incidence among black children in the U.S. and in Africa, and among Arab and Indian children (Kasisi et al, 1990; Linet and Devesa, 1991; Parkin et al., 1992).
Chronic lymphocytic leukemia is almost nonexistent before age 30, then increases rapidly with age, except among Asians older than 50 (Finch and Linet, 1992; Parkin et al., 1992).
Acute myelocitic leukemia is the subtype of highest incidence among young and middle-aged adults, with rates consistently higher in more developed countries and in urban areas (Cartwright and Staines, 1992; Parkin et al., 1992).
Chronic myelocytic leukemia accounts for 1 to 3 percent of childhood leukemia, rises in adolescence, then increases more rapidly in early adulthood, although rates are lower than those for AML. Rates for all types of leukemia are higher among males than among females (Parkin et al., 1992), and among Caucasians than blacks, except for CML (Groves et al, in press).
Childhood leukemia death rates have dramatically decreased since the 1960s because of treatment advances (Linet and Devesa, 1991; Aoki et al., 1992). Increases in CLL among the elderly within the past few decades have been attributed to improvements in diagnosis (Finch and Linet, 1992), whereas increases in AML among men 50 years and older in industrialized regions may reflect occupational exposures (Sandler and Collman, 1987).
Numerous families have been described in which two or more closely related members have been diagnosed with leukemia or related blood malignancy (Linet, 1985; Finch and Linet, 1992). In several large series of leukemia patients, 5 to 10 percent have reported additional close relatives with leukemia or related blood malignancies compared with 1 to 2 percent of similarly affected families of healthy persons (Pottern et al., 1991). Children with Down's syndrome or other abnormal chromosome condition are at increased risk of developing leukemia (Robison and Neglia, 1987). Most acute leukemia cases have been found to have chromosomal abnormalities (Sandler and Collman, 1987).
The genetic or environmental factors responsible for familial occurrence or chromosome changes in individual leukemia patients are unknown, although chromosome abnormalities in a few cases have been linked with exposure to ionizing (X-rays or gamma rays) radiation, benzene, other solvents, pesticides, or to treatment with certain types of chemotherapy drugs (Sandler and Collman, 1987).
Excessive leukemia deaths were reported among early U.S. (Matanoski et al., 1984) and British radiologists (Smith and Doll, 1981) with high radiation exposures, but, because of lower doses and the use of shielding, no excess has occurred for several decades.
Most studies of nuclear industry or shipyard workers have shown either no leukemia excess or small nonsignificant increases (National Research Council, 1990; Kendall et al., 1992). Some studies of nuclear industry workers have shown a positive dose-response relationship (Wilkinson and Dreyer, 1991; Kendall et al., 1992), while others have shown no association between radiation exposure level and leukemia risk (Fraser et al., 1993). Leukemia mortality was elevated among British (Darby et al., 1988) and New Zealand military (Pearce et al., 1990), but not among most U.S. servicemen (Robinette et al., 1985), participating in atmospheric nuclear tests.
Japanese children and adults exposed to high radiation levels experienced an increased leukemia risk (except for CLL) that peaked about five years subsequent to the atomic bomb blasts in Hiroshima and Nagasaki (National Research, 1990; Preston et al., 1994). Children born to women pregnant at the time of the blasts in Japan did not develop an elevated occurrence of leukemia, although an excess has been observed among children born and residing in proximity to Sellafield nuclear reprocessing plant in Great Britain despite very low measured radiation levels; the excess was observed among children and persons under age 25 before 1983 but not after that year (Draper et al., 1993).
Some studies have shown leukemia increases among children whose fathers were employed in the nuclear industry (Gardner et al., 1990; Roman et al., 1993), but others have reported no increases (Kinlen et al., 1993). Numerous studies in Utah of a possible relationship of leukemia with fallout from nuclear weapons have been conflicting (Machado et al, 1987; Stevens et al., 1990). No increase in childhood or adult leukemia mortality was found in 113 U.S. counties adjacent to 62 nuclear plants compared with death rates in control counties (Jablon et al., 1991), nor have childhood leukemia excesses been identified to date in studies of cancer registries in European countries subsequent to the accident in 1986 at the nuclear plant in Chernobyl in the former Soviet Union (Parkin et al., 1993; Auvinen et al., 1994; Hjalmars et al., 1994).
A 50 percent increase in childhood leukemia has generally been associated with pregnancy-related diagnostic X-ray exposure (MacMahon, 1962), but a few studies in children and studies examining the relationship of diagnostic X-ray exposure in adults with leukemia have been conflicting (Gibson et al., 1976; Boice et al., 1991). Radiotherapy treatments have been associated with leukemia excess among patients with ankylosing spondylitis (Darby et al., 1987), and with small increases among women with cervical and uterine cancer (Boice et al., 1987), heavy menstrual bleeding due to benign conditions (Inskip et al., 1990), breast cancer (Curtis et al., 1992), and Hodgkin's disease (Tucker et al., 1988), although splenectomy may play a role in the latter (Tura et al., 1993).
Children radiated for fungal infection of the scalp (Ron et al., 1988) or for large thymuses in infancy (Hildreth et al., 1985) have an elevated risk, although children treated for cancer with radiotherapy alone have not been found to develop increased leukemia (Tucker et al., 1987).
Electricians, power line workers, and electronics and other workers thought to be exposed to nonionizing electrical and magnetic fields have been reported to have an elevated leukemia (primarily AML) risk in some but not all studies in which assessment of occupational exposure was based on job titles (Savitz and Calle, 1987).
A large study of Canadian and French utility workers demonstrated a small excess of AML (Theriault et al., 1994). Childhood leukemia has also been weakly linked with proxy measures of residential exposure to magnetic fields in some, but not all, studies (National Radiological Protection Board, 1992).
Benzene-exposed shoe, leather, rubber, and chemical manufacturing workers have been repeatedly shown to have excess leukemia (primarily AML that is two- to ten-fold increased) (IARC, 1981). Elevated leukemia risks have also occurred among some rubber manufacturing (Delzell and Monson, 1981), petroleum refinery, and chemical plant workers (Wong and Raabe, 1989), and some pressmen and printers (Paganini-Hill, 1980), painters (Matanoski et al., 1986), and various other occupational groups.
Small excesses have also been found among farmers in some regions; suspect exposures include certain livestock (e.g., viruses associated with poultry and dairy cows), pesticides, and other agrochemicals (Blair et al., 1992).
Childhood leukemia has been linked in some studies with parental hydrocarbon-related, chemical or metal manufacturing, and other occupational exposures (Savitz and Chen, 1990). Interview data collected on environmental pesticide exposures after birth have also been associated with childhood leukemia in two studies (Lowengart et al., 1987; Shu et al., 1988), and are currently being evaluated as causes of leukemia in the first investigations with measurements of residential pesticide levels.
A possible association between cigarette smoking and adult leukemia was first suggested in 1986. Since then, follow-up studies in different U.S. populations have shown elevated risk of leukemia, particularly myeloid leukemia (Brownson et al., 1993; Siegel, 1993). Some studies comparing acute myeloid leukemia cases with controls have also shown this link (Brownson et al., 1993), although others and an earlier follow-up investigation of British doctors (Doll et al., 1976) have shown no excess of AML among smokers.
Despite considerable interest for decades in apparent leukemia clusters, results of multiple investigations have been inconclusive (Linet, 1985; Cartwright and Staines, 1992). Viruses were long suspected (because of their known role in causing leukemia and related blood malignancies in animals), but the first human leukemia virus (a retrovirus called HTLV-I) was not identified until 1980. Since then, HTLV-I has been closely linked with a rare adult T-cell leukemia that clusters in southern Japan, the Caribbean, parts of Africa, and in immigrants from these regions to the United States. Approximately 2 to 12 percent of healthy persons in these areas show evidence of viral infection; among the infected individuals, lifetime risk of adult T-cell leukemia is 1 to 4 percent. The virus is spread through transfusion, by intravenous drug abuse, sexual intercourse, and breast-feeding (Blattner, 1993).
Leukemia has also been inconclusively linked with several immune-related diseases (Cartwright and Staines, 1992; Finch and Linet, 1992). It is often not clear whether the associations of leukemia with previous nonmalignant diseases are due to the illness, to increased medical attention (resulting in increased diagnostic X-ray exposure and/or earlier diagnosis), or to drug treatments used for that condition. Unfortunately, AML has been consistently linked with a class of effective, important chemotherapy drugs called alkylating agents, which have been successfully used to treat many types of cancer as well as certain nonmalignant conditions (Pedersen-Bjergaard and Philip, 1989).
Despite the known leukemia-causing effect of these drugs, equally effective alternatives for treating certain life-threatening cancers have yet to be identified. A few reports have also suggested that chloramphenicol (Shu et al., 1988), growth hormone (Fradkin et al., 1993) and others may be associated with leukemia, but definitive evidence is lacking.
Although numerous epidemiologic studies have assessed possible risk factors for the leukemias, the etiology of most cases is largely unknown. To date, lifestyle, diet, and most residential environmental exposures have received little attention.