Genetic Susceptibility to Lung Cancer
Christopher I. Amos
Joan E. Bailey-Wilson
Lung cancer is the most common cause of cancer death in the United States1 with 162,000 deaths estimated for 2008 among 215,000 incident cases. Lung cancer became the leading cause of cancer death in men in the early 1950s and in women in 1987. From 1950 to 1988, lung cancer experienced the largest increase in mortality rate of all the cancers,2 reflecting increases in smoking behavior. Lung cancer prognosis often remains poor because it is usually detected as a stage III or IV malignancy. Identifying genetic factors that influence lung cancer risk could help in identifying subsets of individuals at particularly high risk, in whom early detection strategies could be adopted. In addition, a better understanding of the genetic influences that increase lung cancer risk may lead to the development of novel approaches for chemoprevention and therapy.
Cancer of the lung has frequently been cited as an example of a malignancy that is solely determined by the environment 3 and the risks associated with cigarette smoking3,4,5,6 and certain occupations, such as mining,7 asbestos exposure, shipbuilding, and petroleum refining,8,9,10,11,12 are well established. About 85% to 90% of lung cancer risk can be associated with cigarette smoking.13,14,15 Environmental tobacco smoke (ETS, passive smoking) has also been shown to be associated with a mild increase in risk for lung cancer in North America and Europe.6,16,17,18 A recent prospective European study estimating that between 16% to 24% of lung cancers in nonsmokers and long-term ex-smokers were attributable to ETS.19 A metaanalysis of 22 studies showed that exposure to workplace ETS increased risk of lung cancer in workers by 24%, and that this risk was highly correlated with duration of exposure.20
Dietary studies have found reduction in risk associated with high compared with low consumption of carotene-containing fruits and vegetables.21,22,23,24,25,26,27 Subsequent chemoprevention trials among high-risk subjects with a history of smoking and/or occupation exposure showed a surprising increase in lung cancer risk in these populations associated with beta-carotene supplementation. 28,29,30 At least one very large metaanalysis31 found significant protective effects of increased levels of dietary β -cryptoxanthin, another carotenoid.
There is little doubt that most lung cancer cases are attributable to (i.e., would not occur in the absence of) cigarette smoking and other behavioral and environmental risk factors,2,6,16,32,33 and most studies indicate that duration of cigarette smoke is a more important risk factor than intensity or number of cigarettes smoked per day.34,35,36 However, it was conjectured long ago that individuals differ in their susceptibility to these environmental insults.37,38,39 Mutations and loss of heterozygosity at genetic loci such as oncogenes and tumor suppressor genes (TSGs) are involved in lung carcinogenesis, 40,41,42 but most of these changes are thought to accumulate in individual somatic cells over time, as opposed to the inherited risk to all cells that will result from mutations or risk-increasing polymorphisms occurring in transmission from germ cells. However, numerous studies show that certain allelic variants at some genetic loci affect inherited susceptibility to lung cancer. Furthermore, mounting epidemiologic evidence has suggested that lung cancer shows familial aggregation after adjusting for cigarette smoking and other risk factors, and that differential susceptibility to lung cancer is inherited in some families. This chapter describes inherited major susceptibility loci and findings from well-replicated studies of loci for lung cancer risk that have less pronounced effects. We also relate these risks to the well-known risks as a result of environmental risk factors, particularly personal cigarette smoking.
INHALATION OF TOBACCO SMOKE
The association between cigarette smoking and lung cancer is strong and well established.3,4,5,43,44,45,46,47 The incidence of lung cancer is correlated with the cumulative amount and duration of cigarettes smoked in a dose-response relationship,6,44,48 and smoking cessation results in a leveling off of risk for lung cancer at the time of smoking cessation.6,49,50 Lung cancer rates and smoking rates are also highly correlated in different geographic regions.51 In 1991, Shopland et al.52 showed that the relative risk (RR) of lung cancer for male smokers versus nonsmokers is
22.36 and that for female smokers versus female nonsmokers is 11.94. They also estimated that 90% of lung cancers in men and 78% in women were directly attributable to tobacco smoking. Kondo et al.53 showed a significant (p <0.001) dose-response relationship between number of cigarettes smoked and the frequency of p53 mutations in tumors of lung cancer patients, suggesting that somatic p53 mutations may be caused by exposure to a carcinogen/mutagen in tobacco smoke or its metabolites.
22.36 and that for female smokers versus female nonsmokers is 11.94. They also estimated that 90% of lung cancers in men and 78% in women were directly attributable to tobacco smoking. Kondo et al.53 showed a significant (p <0.001) dose-response relationship between number of cigarettes smoked and the frequency of p53 mutations in tumors of lung cancer patients, suggesting that somatic p53 mutations may be caused by exposure to a carcinogen/mutagen in tobacco smoke or its metabolites.
BIOLOGIC RISK FACTORS
In general, all studies suggesting genetic susceptibility have also shown strong risk resulting from cigarette smoking and often have shown an interaction of high-risk genotype and smoking on lung cancer risk. When trying to determine whether a complex disease or trait such as lung cancer has a genetic susceptibility, one asks three major questions:
Does the disease (lung cancer) cluster in families? If some risk for lung cancer is inherited, then one would expect to see clustering of that cancer in some families above what would be expected by chance.
If the aggregation of lung cancer does occur in some families, can the observation be explained by shared environmental/cultural risk factors? In the study of lung cancer, one needs to assess whether the familial clustering of lung cancer is solely a result of clustering of smoking behaviors or other environmental exposures within families.
If the excess clustering in families is not explained by measured environmental risk factors, is the pattern of disease consistent with Mendelian transmission of a major gene (i.e., of transmission through some families of a moderately high penetrance risk allele) and can this gene(s) be localized and identified in the human genome?
In addition, inherited susceptibility factors for lung cancer can also be identified by conducting large-scale case control studies, just as these approaches have been successful for other complex diseases that often result from a complex interplay of genetic and environmental factors.
EVIDENCE FOR FAMILIAL AGGREGATION OF LUNG CANCER
Epidemiologic Cohort Studies Tokuhata and Lilienfeld 54,55 showed familial aggregation of lung cancer over 40 years ago. After accounting for personal smoking, their results suggested the possible interaction of genes, shared environment, and common lifestyle factors in the etiology of lung cancer. In their study of 270 lung cancer patients and 270 age-, sex-, race-, and location-matched controls and their relatives, they found an RR of 2.0 to 2.5 for mortality because of lung cancer in cigarette-smoking relatives of cases compared with smoking relatives of controls. Nonsmoking relatives of lung cancer cases were also at higher risk when compared with nonsmoking relatives of controls. Smoking was a more important risk factor for men, but family history was the more important risk factor for women. They also noted a synergistic interaction between familial and smoking factors on the risk of lung cancer in relatives, with smoking relatives of lung cancer patients having much higher risk of lung cancer than either nonsmoking relatives of patients or smoking relatives of controls. They observed a substantial increase in mortality resulting from noncancerous respiratory diseases in relatives of patients compared with relatives of controls, suggesting that the case relatives have a common susceptibility to respiratory diseases. However, they found no significant differences between the spouses of the lung cancer cases and controls for lung cancer mortality, mortality from noncancerous respiratory diseases, or smoking habits.
The major weakness of this study was that smoking status alone was used. Therefore, some of the familial aggregation could be a result of familial correlation in smoking levels or age at smoking initiation. However, nonsmoking relatives of cases were at higher risk than nonsmoking relatives of controls.
At present, many other studies have shown evidence of familial aggregation of lung cancer. In 1975, Fraumeni et al.56 reported an increased risk of lung cancer mortality in siblings of lung cancer probands. In 1982, Goffman et al.57 reported families with excess lung cancer of diverse histologic types. Lynch et al.58 reported evidence for increased risk of cancer at all anatomic sites for relatives of lung cancer patients but no significant increased risk for lung cancer alone in these relatives.
In southern Louisiana, case-control studies reported an increased familial risk for lung cancer59 and smoking-related non-lung cancers60 among relatives of lung cancer probands (the index case leading the family to be studied) after allowing for the effects of age, sex, occupation, and smoking. In these two studies, familial aggregation analyses were performed on a set of 337 lung cancer probands (cases), their spouse controls, and the parents, siblings, half-siblings, and offspring of both the probands and the controls. The probands were male and female whites who died from lung cancer during the period 1976 to 1979 in a 10-parish (county) area of southern Louisiana, a region noted for its high lung cancer mortality rates. There were about 3.5 male probands to every female lung cancer proband in the data set. A strong excess risk for lung cancer was detected among first-degree relatives of probands compared with relatives of spouse controls, after adjusting for age, sex, smoking status, total duration of smoking, cigarette pack-years, and a cumulative index of occupational/industrial exposures. Parents of probands had a fourfold risk of having developed lung cancer compared with parents of spouses, after adjusting for the effects of age, sex, smoking, and occupational exposures. Women greater than 40 years old who were relatives of probands were at nine times higher risk than similar female relatives of spouses, even among nonsmokers who had not reported excessive exposure to hazardous occupations. Among female heavy smokers who were relatives of probands, the risk was increased from fourfold to sixfold. Overall, male relatives of probands had a greater risk of lung cancer than their female counterparts. After controlling for the confounding effects
of the measured environmental risk factors, relationship to a proband remained a significant determinant of lung cancer, with a 2.4 odds in favor of relatives of probands.
of the measured environmental risk factors, relationship to a proband remained a significant determinant of lung cancer, with a 2.4 odds in favor of relatives of probands.
These same families were reanalyzed60 to determine if non-lung cancers exhibited similar familial aggregation. When analyzing the number of cancers at any site that occurred in a family, proband families were found to be 1.7 times more likely than spouse families to have one family member (other than the proband) with cancer, and 2.2 times more likely to have two family members with cancer. Comparing case relatives and control relatives, families that had three and four or more cancers occurred with RRs of 3.7 and 5.0, respectively. Each risk estimate was significant at the 0.01 level. The most striking differences in cancer prevalence between proband and control families were noted for cancer of the nasal cavity/sinus, mid-ear, and larynx (odds ratio [OR] = 4.6); trachea, bronchus, and lung (OR = 3.0); skin (OR = 2.8); and uterus, placenta, ovary, and other female organs (OR = 2.1). After controlling for age, sex, cigarette smoking, and occupational/industrial exposures, relatives of lung cancer probands maintained an increased risk of non-lung cancer (p <0.05) when compared with relatives of spouse controls.
A family case-control study, drawn from a population-based registry in Saskatchewan, Canada was reported by McDuffie.61 A total of 359 cases and 234 age-and gender-matched community controls were included in the study. Most families reported at least one member with a history of neoplastic disease exclusive of the proband (62% of patients’ families and 57% of control families). However, the families of the lung cancer cases were more likely (30%) to have two or more family members affected with any cancer than the families of the controls. The case families were also significantly more likely to have two or more relatives with lung cancer than were the control families. In addition, a higher percentage of all primary tumors were lung tumors (16.5%) in patients’ relatives compared with controls’ relatives (10%). The progression of increased risks for observing 1, 2, 3, and 4+ affected relatives in case families versus control families was less than that observed by Sellers et al.’s60 study but showed the same type of progression.
Family history data from an incident case-control study in Texas were analyzed for evidence of familial aggregation by Shaw et al.62 A total of 943 histologically confirmed lung cancer cases and 955 age-, gender-, vital status-, and ethnicitymatched controls were interviewed regarding smoking, alcohol use, cancer in first-degree relatives, medical history, and demographic characteristics. After adjusting for personal smoking status, passive smoking exposure (ever/never), and gender, there was a 1.8-fold OR associating lung cancer with having one or more first-degree relatives with lung cancer. Lung cancer risk increased as the number of relatives with cancer increased and was highest when only relatives with lung cancer were considered (ORs of 1.7 and 2.8 for one and two or more relatives with lung cancer, respectively). Lung cancer was diagnosed at a significantly younger age among cases who had first-degree relatives with lung cancer than among those who had no relatives with lung cancer. However, no such age difference was seen between cases who had first-degree relatives with any cancer versus those who had no relatives with cancer. This study also examined histologic subtypes of lung cancer cases and found that for each histologic type, there were significant risks associated with having any relatives with lung cancer, with ORs of 2.1 for adenocarcinoma, 1.9 for squamous cell carcinoma, and 1.7 for small cell lung cancer. Finally, in this study, only current and former smokers had an increased lung cancer risk associated with lung cancer in relatives.
Cannon-Albright et al.63 examined the degree of relatedness of all pairs of lung cancer patients in the Utah Population Database (UPD). By comparing this with the degree of relatedness in sets of matched controls, they showed that lung cancer exhibited excess familiality, and three of four histological tumor types still showed excess familiality when considered separately. In the same population, but using different methodology, Goldgar et al.64 studied lung cancer probands and controls who had died in Utah and their first-degree relatives. They found that 2.55 times more lung cancers occurred in first-degree relatives of lung cancer probands than expected (computing a familial relative risk or FRR) based on rates in control relatives. When they stratified by gender, they observed higher relative risks for female relatives of female probands (FRR = 4.02) versus male relatives of male probands (FRR = 2.5). No adjustment was made in these analyses for personal smoking or other environmental risk factors, so these results may partly reflect the familiality of smoking behaviors. However, the UPD is derived from the Church of Latter-Day Saints records, which is largely a nonsmoking population and Utah has the lowest smoking rates of any state in the United States.
In 2000, Bromen et al.,65 in a population-based case-control study in Germany, showed that lung cancer in parents or siblings was significantly associated with an increased risk of lung cancer and that this risk was much stronger in younger participants. In 2003, Etzel et al.66 evaluated whether first-degree relatives of lung cancer cases were at increased risk for lung cancer and for other smoking-related cancers (bladder, head and neck, kidney, and pancreas). They studied 806 hospital-based lung cancer patients and 663 controls matched on age, sex, ethnicity, and smoking history, all from the Houston, Texas area. After adjustment for smoking history of patients and their relatives, there was significant evidence for familial aggregation of lung cancer and of smoking-related cancers. However, they did not find increased aggregation in the families of young onset (less than or equal to age 55) lung cancer cases or in families of never-smokers.
Two studies in China67,68 both found, after adjusting for age, sex, birth order, residence, family size, chronic obstructive pulmonary disease (COPD), smoking, and cumulative index of smoky coal exposure or occupational/industrial exposure index, that first-degree relatives of lung cancer patients were at significantly increased risk for lung cancer compared with the same relatives of controls. They also observed that families of the lung cancer patients were significantly more likely to have three or more affected relatives than were control families.
A series of studies using the Swedish Family-Cancer Database,69,70,71,72 which totals over 10.2 million individuals,
found that a high proportion of lung cancers diagnosed before the age of 50 appear to be heritable, and that lung cancer patients with a family history of lung cancer were at a significantly increased risk of subsequent primary lung cancers. A recent study73 utilizing the Icelandic Cancer Registry calculated risk ratios of lung cancer in first-, second-, and third-degree relatives of 2756 lung cancer patients diagnosed between 1955 and 2002. RRs were significantly elevated for all three classes of relatives, and this increased risk was stronger in relatives of early onset lung cancer patients (age at onset less than or equal to 60 years). The effect did not appear to be solely a result of the effects of smoking in all relative types, except for cousins and spouses.
found that a high proportion of lung cancers diagnosed before the age of 50 appear to be heritable, and that lung cancer patients with a family history of lung cancer were at a significantly increased risk of subsequent primary lung cancers. A recent study73 utilizing the Icelandic Cancer Registry calculated risk ratios of lung cancer in first-, second-, and third-degree relatives of 2756 lung cancer patients diagnosed between 1955 and 2002. RRs were significantly elevated for all three classes of relatives, and this increased risk was stronger in relatives of early onset lung cancer patients (age at onset less than or equal to 60 years). The effect did not appear to be solely a result of the effects of smoking in all relative types, except for cousins and spouses.