Familial lung cancer

Abbreviations: CI = confidence interval; CYP1A1 = cytochrome P4501A1; GSTM1 = glutathione S-transferase M1; MPO = myeloperoxidase; OR = odds ratio

In 2003, an estimated 171,900 new diagnoses and 157,200 deaths from lung cancer occurred in the United States. (1) While 80 to 90% of lung cancer incidence can be attributed to cigarette smoking, only 10 to 15% of all smokers develop lung cancer, (2) and 10 to 15% of all lung cancers occur among nonsmokers. These figures suggest that there are individual differences in susceptibility to lung carcinogens and that these individual differences in susceptibility may be the result of genetic predisposition to lung cancer.

The hallmarks of an inherited disease include familial aggregation above that related to shared environmental or cultural exposures. Once familial aggregation is established, the pattern of inheritance of genetic susceptibility can be evaluated using segregation analysis to determine whether there is statistical evidence fi3r the inheritance of a Mendelian major gene. If there is evidence supporting an inherited component to disease, the localization of disease genes is pursued.

FAMILIAL AGGREGATION OF LUNG CANCER

The most definitive studies of the familial risk of lung cancer have included the collection of smoking and other risk factor information for all first-degree relatives. Four more recent studies (3-6) of familial aggregation of lung cancer have been performed using a detailed family study approach, collecting risk factor data for family members. In a Louisiana family study, (3) after adjusting for age, sex, smoking history, and occupational exposures for each relative, a 2.4-fold excess of lung cancer was reported among relatives of lung cancer patients compared to relatives of spouse control subjects. In Detroit, a study was designed to focus on individuals who were likely to have a risk of lung cancer that was most strongly associated with genetic factors (ie, early-onset cases diagnosed < 49 years of age) and individuals who demonstrated increased susceptibility to low levels of exposure4 (ie, nonsmoking and early-onset cases). An excess risk of lung cancer, after adjustment for each relative's age, sex, race, smoking status, occupation, industry, and history of other lung diseases, was seen in first-degree relatives of nonsmoking lung cancer cases diagnosed 40 to 59 years of age (relative risk, 6.1; 95% confidence interval [CI], 1.1 to 33.4)4 and relatives of patients with early-onset cases (relative risk, 2.5; 95% CI, 1.7 to 3.6) [unpublished preliminary findings]. Familial risk was greater in African Americans (odds ratio [OR], 4.5; 95% CI 2.2-9.11 than in whites (OR, 1.8; 95% CI, 1.2 to 2.9). Mayne et al (5) also studied familial risk in relatives of nonsmoking men and women, and found elevated lung cancer risk in fathers, however, this finding was not significant (OR, 1.85; 95% CI, 0.80 to 4.33). A similar increased risk was reported from a case-control family study (6) in Germany that was not limited to nonsmokers (OR, 1.67; 95% CI, 1.11 to 2.52).

These findings suggest a role for a genetic predisposition to lung cancer after taking into account the familial clustering of smoking habits, family size, and age structure. The findings of stronger aggregation when the onset of disease is early are indicative of an inherited component to risk.

MENDELIAN INHERITANCE OF LUNG CANCER

Studies of familial aggregation do not explicitly reveal possible modes of inheritance underlying familial aggregation. Only two studies (7,8) have investigated whether there is statistical evidence for the inheritance of a major gene for lung cancer using segregation analysis. Both studies (7,8) have reported that the pattern of lung cancer occurrence in families is consistent with mendelian codominant inheritance of a rare autosomal gene. Sellers et al (7) estimated that this putative gene is responsible for 69% of the lung cancer seen at age 50 years, 47% at age 60 years, and 22% at age 70 years. Yang et al (9) reported that an environmental model with homogeneous risk across generations best explained the observed data in families of nonsmokers. However, among families of nonsmoking patients under the age of 60 years, a mendelian codominant model, with significant modifying effects of smoking and chronic bronchitis, best explained the observed data, with an estimated risk allele frequency of 0.004. (8) Homozygous individuals with the risk allele are rare in the study population (1.6 per 100,000 population), making the attributable risk very low, even in the face of the very high penetrance of early-onset lung cancer (men, 85% by age 60 years; and women, 74% by age 60 years). One percent of the study population was made up of heterozygous individuals who had a relatively low risk of lung cancer, unless they were smokers with chronic bronchitis.

GENES FOR LUNG CANCER

Findings of familial aggregation and statistical evidence for a major gene have led to the search for high-penetrant, rare, single genes for lung cancer and low-penetrant, high-frequency, susceptibility genes for lung cancer.

High-Penetrant, Low-Frequency Genes for Lung Cancer

A single gene for lung cancer has not yet been identified, although lung cancer does occur, on occasion, in families with Li-Fraumeni syndrome, a condition that is associated with inherited p53 mutations. (10) Large, multi-generation pedigrees with multiple affected family members need to be accrued for genome-wide searches for lung cancer genes. The difficulties in pursuing a single gene for lung cancer are that lung cancer families are rare (occurring in only 1% of the population), onset is usually in the mid to late 60s, and, because of the high case fatality for lung cancer, affected relatives are typically deceased, and smoking data must be collected on all family members, many of whom also may be deceased. These problems necessitate collaborative efforts to identify families for linkage studies. One such collaborative effort is underway by the Genetic Epidemiology of Lung Cancer Consortium.

High-Frequency, Low-Penetrant Genes for Lung Cancer

Genes encoding enzymes that are associated with carcinogen metabolism and DNA repair have been the focus of research into possible susceptibility genes for lung cancer. It has been hypothesized that differences in susceptibility to carcinogens result from an individual's ability to form genotoxic intermediates, to detoxify these intermediates, and to repair damaged DNA. Polymorphisms in genes coding for the enzymes that drive these processes are likely candidate susceptibility genes.

Some of the most widely and recently studied polymorphic loci coding for phase I and II enzymes involved in the activation and conjugation of tobacco smoke constituents are cytochrome P4501AI (CYP1A1), glutathione S-transferase M1 (GSTM1), myeloperoxidase (MPO), and NAD(P)H: quione oxidoreductase (NQ01). The results for selected larger studies are presented in Table 1. The findings of an association between CYP1A1 polymorphisms and a risk of lung cancer have been inconsistent. Studies in Japanese populations (11,12) have reported risks that were increased over twofold. A meta-analysis (13) based on 15 studies reported a nonsignificant OR of 1.27 associated with the MspI polymorphism and a nonsignificant OR of 1.62 for the exon 7 polymorphism in CYP1A1. In pooled analyses based on 22 studies, a significant 2.4-fold increase was reported for the homozygote MspI variant, however, this finding was based on very small numbers of cases and controls carrying the risk genotype. (14)

The gene encoding glutathione S-transferase mu (GSTM1) occurs in the null form in approximately half of the population. One meta-analysis (15) estimated an overall OR of 1.13 (95% CI, 1.04 to 1.25), suggesting a modest increased risk associated with the GSTM1 null genotype. An updated meta-analysis (16) reported similar findings (OR, 1.17; 95% CI, 1.07 to 1.27). A pooled analysis (17) in whites < 45 years of age also found an OR of 1.1 (95% CI, 0.9 to 1.3), while a study (18) on an African-American population reported a twofold increased risk (95% CI, 1.07 to 4.11) among those persons with the null genotype. Analyses (19-21) of combined CYP1A1 and GSTM1 risk genotypes have generally showed risk increases of more than threefold, however, sample sizes have usually been small.