Ethnic Differences in Cancer Incidence: A Marker for Inherited Susceptibility?

Frank D. Gilliland

Department of Internal Medicine and Epidemiology, University of New Mexico School of Medicine, Albuquerque, New Mexico

• Introduction
• Explanation for Ethnic Differences in Cancer Occurrence
• Methods
• Results and Discussion

Key words: ethnicity, cancer, molecular epidemiology, susceptibility, BRCA1, p53

This paper is based on a presentation at the symposium on Mechanisms and Prevention of Environmentally Caused Cancers held 21-25 October 1995 in Santa Fe, New Mexico. Manuscript received at EHP 16 April 1996; accepted 16 July 1996.

Supported by Contract NO1-CN-05228 from the Division of Cancer Prevention and Control, National Cancer Institute.

Address correspondence to Dr. F.D. Gilliland, Department of Internal Medicine and Epidemiology and Cancer Control Program, University of New Mexico School of Medicine, 900 Camino De Salud NE, Albuquerque, NM 87131. Telephone: (505) 277-5541. Fax: (505) 277-8572. E-mail: fgillila@medusa.unm.edu

Abbreviations used: SEER, Surveillance, Epidemiology, and End Results Program; SES, socioeconomic status

Introduction

Explanation for Ethnic Differences in Cancer Occurrence

Measurement errors:

• Inadequate data--insufficient information, based upon clinical impressions, etc

• Differential access to medical care and diagnostic facilities

• Differences in reporting due to cultural factors or to difference in the severity of disease; differential use of available facilities

• Differing fashions of diagnosis.

• Coding death certification

• Differences in socioeconomic class and occupation, and secondary factors associated with these differences (see Differences in environment, below)

• Climatic differences and their effects

• Geographic variation in disease frequency

• Nutrition or diet

• Differences in personal customs or habits (e.g., reproductive and nursing habits; use of tobacco and alcohol, and differences in sexual practices)

• Differences with respect to social and family structure relationships, role behavior

• Cultural factors

• Differences related to rates of growth and development

• HLA class II alleles

• HLA-haplotypes

• Metabolic enzyme polymorphisms

• ABO blood groups

Valid comparison of rates depends upon accurate diagnosis and reporting of cancer cases. Bias from measurement errors can result from differences in access to medical care and utilization of care and to differences in diagnosis and death certificate reporting, all of which probably account for a portion of the ethnic variation in cancer rates (2). The bias from measurement error is likely to be substantial and may also explain some of the international variation in cancer mortality rates. However, international standardization of registration procedures has resulted in improved data on cancer incidence worldwide (2), and it is doubtful that information bias explains much of the ethnic variation in rates calculated from data collected by the Surveillance, Epidemiology, and End Results Program (SEER) in the United States.

There are marked differences in the United States and worldwide in the ethnic distribution of age, socioeconomic status (SES), and occupation (3,7-11). Because these factors are strong determinants of cancer risk, differences in their distribution among ethnic groups could explain a portion of the ethnic variation in cancer rates. However, variation in age is accounted for in the data by age adjustment and does not contribute to ethnic differences in rates; while critical review of the literature is beyond the scope of this discussion, SES and occupation do not appear to fully explain the ethnic variation in cancer risk for most cancer sites (3,7,9-11).

Environmental factors, including lifestyle and dietary factors, have traditionally been thought to be the main contributors to ethnic differences in cancer rates (3,9,11). In a now classic analysis, Doll and Peto (11) examined international differences in cancer rates and concluded that 80% of cancers had environmental causes. However, new biological evidence and theories of carcinogenesis suggest this interpretation of ethnic variation in cancer rates may need to be revised.

Carcinogenesis is now recognized to be a multistep process with crucial steps involving mutations or heritable changes in expression of key genes involved in cellular growth control and genome stability (12-14). Cancer risk is determined by the probability of mutations in key genes (13). In this model, individual genetic susceptibility can arise by two pathways; first, from mutations in key genes on the pathway to cancer, such as oncogenes and tumor suppressor genes; and second, from genotypes that increase the probability of mutations in key genes in conjunction with specific environmental exposures.

The role of genetic susceptibility as an explanation for ethnic variation in cancer risk has not been extensively studied. Overall measures of genetic differences in populations indicate that genetic variation is as least as great within ethnic groups as between groups. However, ethnic variation in the distribution of specific susceptibility genotypes does occur. The distribution of a number of mutations in tumor suppressor genes and metabolic polymorphisms has been reported to vary by ethnic group (15-17).

Incidence rates for breast cancer show marked contrasts among ethnic groups (16,18-20). Mutations in the BRCA1 gene are associated with increased risk for breast and ovarian cancer. This gene is composed of 5592 nucleotides spread over 100,000 bases of genomic DNA. It contains 22 coding exons that produce an 1863 amino acid protein, which shows no homology to any known protein except for a RING finger motif near the N-terminus. It is thought to act as a tumor suppressor gene (21).

Carriers of BRCA1 mutations are heterozygotes and have been shown to have a greater than 85% lifetime risk of developing breast cancer and 45% risk of ovarian cancer compared to a 12% risk for women in the general popualtion (22,23). The risk of breast cancer for carriers of BRCA1 mutations varies by age; women 50 years of age have a 50% risk for breast cancer. The frequency of BRCA1 mutations within a population varies between ethnic groups, from 1 in >1000 for Japanese to 1 in 100 for Ashkenazi Jews (16,19,24). Studies have also indicated that some mutations are specific for a given ethnic group, such as the 185 delAG mutation found in the Ashkenazis (25). Differences in genotype distribution may result from differences in consanguinity, mutation rate, natural selection, and random effects such as founder effects and isolation (4).

To consider the potential contribution of genetic susceptibility to ethnic variation in cancer incidence, we used simple probability models to estimate the magnitude of cancer risk differences that might stem from ethnic differences in genetic susceptibility arising from one of the two pathways to increased risk and inheritance of mutations in a tumor suppressor gene. We assumed the simple case where risk is independent of exposure.

Methods

The proportion of the population with susceptibility genotypes depends upon whether the susceptible allele, S, is dominant or recessive. If it is dominant, as with tumor suppressor genes, both SS and NS genotypes will be susceptible and the proportion of susceptibles in the population will be given by q(2-q), where q is the susceptible allele frequency. For the case where the susceptibility allele is recessive, only the SS genotype will be susceptible and the proportion of susceptibles in the population will be given by q². For a susceptibility allele frequency of 10%, a dominant susceptibility allele will result in 19% being susceptible. Under a recessive model, only 1% of the population is susceptible. In the following models, the susceptible proportion will be used as the parameter for population genetic susceptibility.

In a comparison of rates in two ethnic groups, where RRe=ethnic relative risk, Ra=the disease risk in ethnic group A, and Rb=the disease risk in ethnic group B

is an accepted measure of ethnic variation in cancer risk.

In the simple case in which cancer risk is determined by inheritance of a mutation in a single tumor suppressor gene and ethnic differences in risk arise from differences in the allele distribution of this gene, the ethnic relative risk can be expressed as a ratio of disease risk between the two ethnic groups:

where Pa and Pb are the proportions of susceptibles in groups A and B, respectively, and Rg is the risk ratio for those with the susceptible genotype compared with those with the nonsusceptible genotype. Assumptions for this model are that baseline risks are equal in the two ethnic groups, and Rg is constant and independent of exposure or mutation spectrum.

Figure 1. Relative risk comparing any two ethnic groups' assymtotic behavior for a fixed ratio of susceptibility genotype proportion.

Figures 1 and 2 illustrate the general form of the relationship among RRe, Rg, and the distribution of the proportion of susceptibles. For specific examples, we chose to examine the ethnic relative risk that could arise from differences in the proportion with cancer susceptibility arising from tumor suppressor genes with different characteristics, p53 and BRCA1. The germline mutation frequency for p53 is low, approximately 10^-5, but the cancer relative risk is high, in the 10⁴ to 10⁵ range (26). For BRCA1, the frequency is approximately 5 per 1000, but it has been found to show ethnic variation (16,17,25). The relative risk associated with BRCA1 varies with age and is approximately 200 in women aged less than 45 years.

Results and Discussion

The rate at which the ethnic relative risk approaches its maximum value as Rg increases depends upon the magnitude of the proportion of susceptibles in the groups. To see this more concretely, consider two scenarios, both with a Pa/Pb ratio of 5, as shown in Figure 1. First, in the high Pa scenario half of group A is susceptible, so Pa=0.5 and one-tenth of group B is susceptible, so Pb=0.1, giving a Pa/Pb ratio of 5. Second, the low Pa scenario, where Pa=0.05 and Pb=0.01, again a Pa/Pb ratio of 5. As the relative risk for susceptibility genotypes increases, the ethnic relative risk increases to its maximum faster for the high Pa than for the low Pa scenario. Thus for a given susceptibility genotype, relative risk and Pa/Pb ratio, the higher the proportion of susceptibles, the higher the ethnic relative risk.

Figure 2. Relative risk comparing any two ethnic groups with differing proportions of susceptibility genotypes.

In consideration of plausible values of Pa and Pb and relative risks for susceptibility genotypes, Figure 2 shows the ethnic relative risk on the y axis and the relative risk for susceptibility genotype on the x axis. The ranges of the relative risks for the genotype and the groups proportion of susceptibles were chosen for plausible values for tumor suppressor genes p53 and BRCA1.

Figure 2 shows a comparison of two ethnic groups with differing BRCA1 mutation frequencies, with RRe for a susceptible proportion of 1 in 100 versus 5 in 1000. The ethnic relative risk increases rapidly to 1.5 for a susceptibility genotype relative risk of 200. These values of the susceptibility proportion and genotype relative risks are in the ballpark for BRCA1 in young women from specific ethnic groups (16,17,25). Differences in BRCA1 frequency could explain ethnic relative risks for breast cancer in the 1.5 to 2 range for young women.

For a population with lower susceptibility proportions, such as that observed for germline p53 mutations, the ethnic relative risk is small for plausible relative risks for susceptible genotypes. These values are in the range observed for several tumor suppressor genes, indicating that these genes are unlikely to explain even small ethnic differences.

In summary, ethnic differences in cancer occurrence may be a marker of differences in genetic susceptibility. For breast cancer, observed differences in the frequency of BRCA1 mutations could account for ethnic differences in rates for young women. However, the magnitude of ethnic relative risk is likely to more strongly reflect differences in the distribution of susceptibility genotypes between groups than the magnitude of the disease risk associated with the genotypes. For many scenarios, the ethnic relative risk arising from differences in susceptibility may be bounded by the ratio of the proportion of susceptible individuals in each group.

26. Knudson A. Antioncogenes and human cancer. Proc Natl Acad Sci USA 90:10914-10921 (1993).