|Home | About | Journals | Submit | Contact Us | Français|
Family history (FH) of lung cancer is an established risk factor for lung cancer, but the modifying effect of smoking in relatives has been rarely examined. Also, the role of FH of non-malignant lung diseases on lung cancer risk is not well known. We examined the role of FH of cancer and FH of non-malignant lung diseases in lung cancer risk, overall, and by personal smoking, FH of smoking, and histology in 1,946 cases and 2,116 population-based controls within the Environment And Genetics in Lung cancer Etiology (EAGLE) study. Odds ratios (ORs) and 95% CI from logistic regression were calculated adjusting for age, gender, residence, education, and cigarette smoking. FH of lung cancer in any family member was associated with increased lung cancer risk (OR = 1.57, 95% CI = 1.25–1.98). The odds associated with fathers’, mothers’ and siblings’ history of lung cancer were 1.41, 2.14, and 1.53, respectively. The associations were generally stronger in never smokers, younger subjects, and for the adenocarcinoma and squamous cell carcinoma subtypes. FH of chronic bronchitis and pneumonia were associated with increased (OR =1.49, 95% CI = 1.23–1.80) and decreased (OR = 0.73, 95% CI = 0.61–0.87) lung cancer risk, respectively. FH of lung cancer and FH of non-malignant lung diseases affected lung cancer risk independently, and did not appear to be modified by FH of smoking.
Although tobacco smoking is unequivocally the major risk factor for lung cancer, there is substantial evidence for a hereditable component in lung cancer risk based on animal models 1, 2, family studies 3, 4, segregation studies 5–7, linkage analysis 8, and candidate gene association/genome-wide association studies in the context of epidemiological research 9–13. Family history of lung cancer has been consistently shown to be associated with increased lung cancer risk among non smokers 7, 14–23, as well as smokers. Although familial aggregation of lung cancer cannot be solely attributed to shared smoking habit 24, for the large majority of lung cancers, those occurring in smokers, it is difficult to disentangle the specific contribution of the genetic background from the shared smoking exposure in familial settings 25. Limited studies have examined the association of family history of lung cancer and lung cancer risk with adjustment for personal smoking and family history of smoking 26–29. Moreover, few studies have examined the risk associated with different histological groups 30–32. To our knowledge, no study has investigated the role of family history of non-malignant lung diseases in relation to lung cancer risk.
We took advantage of the rich dataset of the Environment And Genetics in Lung cancer Etiology (EAGLE) study 33, a large population-based case-control study of lung cancer, to examine the risk of lung cancer in association with several components of familial aggregation. Specifically, we examined the association of family history of lung cancer, family history of smoking-related tumors, and family history of non-malignant lung diseases in relation to lung cancer risk overall, by histology, by smoking status, and family history of smoking in 2,101 lung cancer cases and 2,120 population controls, taking into account the major risk factors for lung cancer.
Environment And Genetics in Lung cancer Etiology (EAGLE) study (http://dceg.cancer.gov/eagle/) is a population-based case-control study of lung cancer, conducted between 2002 and 2005. Study design and subjects’ characteristics are detailed in Landi et al 2008 33. Briefly, the study included 2,101 incident lung cancer cases, both males and females of Italian nationality, ages 35 to 79 years old, with verified lung cancer of any histological type, enrolled from 13 hospitals within the Lombardy region of Italy. Population-based controls (N=2,120) were randomly sampled from the same catchment area of the lung cancer cases, and frequency-matched to cases by age, gender, and residence. The study benefited from a high participation rate from both cases (86.6%) and controls (72.4%). The study protocol was approved by each Hospital’s and by the National Cancer Institute’s Institutional Review Board and each subject signed an informed consent prior to participation in the study.
Data on demographic characteristics, detailed personal smoking history, family history of cancer in first degree relatives (mother, father, siblings, and offspring), family history of non-malignant lung diseases (chronic bronchitis, emphysema, pneumonia, and tuberculosis) in first degree relatives, and family history of smoking in first degree relatives (whether the particular family member smoked over 100 cigarettes during his/her lifetime) were collected through an interview-based computer-assisted questionnaire from both cases and controls. All data are stored in a secure relational database. Quality control procedures were implemented to ensure accuracy, completeness, and privacy of the data collected.
Odds ratios (ORs) and 95% confidence intervals (CIs) from unconditional logistic regression were calculated to quantify the association between family history of cancer or non-malignant lung diseases and lung cancer risk. The number of siblings in the families ranged from 0 to 18, with a median of 3; 10% of both cases and controls had 7 or more siblings. Since there were only 7 cases and 8 controls with more than one sibling with lung cancer, we defined the family history in siblings, “yes”, as having any affected sibling in the family regardless of the number of affected individuals, and “no” as having no affected sibling in the family. The families that had any sibling with unknown disease status and no sibling with “yes” were excluded from the analysis (486 cases and 431 controls for the family history of lung cancer, and similar distribution between cases and controls for the other diseases). Family history in any family member was defined similarly as there were only 15 cases and 22 controls with more than one first-degree relative affected. Overall, 510 cases and 473 controls were excluded for the analysis of family history of lung cancer in any family member, and similar number of exclusions was applied for the other diseases. We also repeated the analyses without exclusions and classifying family members with unknown family history status as negative (family history “no”). The results were not substantially changed (Supplemental Table 1); we report here the results based on the more restrictive definition with the exclusions of those with unknown family history status.
We explored associations with lung cancer risk overall and stratified by smoking status, or family history of smoking. We conducted two analyses. One model was adjusted only for the matching variables (age, gender and residence area); the second was adjusted for the matching variables, personal cigarette smoking exposure (never, former and current smoking status, intensity (packs/day), duration (years), and years from the last cigarette smoked for former smokers, and education (proxy of socioeconomic status). We observed no large differences between the two models. Results from the full adjusted analyses are reported in the manuscript. The analyses in never smokers were further adjusted for passive smoking exposure (any exposure at home during childhood and adulthood, as well as at the work place). In addition, the analyses of the association between family history of non-malignant lung diseases and lung cancer risk were further adjusted for family history of lung cancer; given the results were not substantially changed, only results without this adjustment were reported. Adjustment for alcohol consumption or for number of siblings in the families did not modify the results and were not included in the final models. When the number of observations in any cell of a 2×2 table was smaller than 5, the OR was not estimated.
We explored the association of family history of lung cancer and family history of non-malignant lung diseases with lung cancer risk by histology. The histology types selected were those with larger number of subjects for the analyses, including squamous cell carcinomas (n=503), adenocarcinomas (n=803), and small cell carcinomas (n=195). Homogeneity among histologic-specific lung cancer risks was assessed using the Wald test. Family history of cancer was stratified by family history of lung cancer, “smoking-related tumors” and “other tumors”. “Smoking-related tumors” included tumors of the larynx, oral cavity and pharynx, esophagus, bladder, kidney, pancreas, cervix and stomach; the “other tumors” included melanoma, leukemia, and cancers of the colon, uterus, small intestine, and brain. Subjects who were not able to define the type or anatomical location of their relatives’ tumors, but could only recall whether they had lung cancer were excluded from the analyses of “smoking-related tumors other than lung cancer” or “other tumors” since it was not possible to accurately classify the relatives’ tumors. P values are two-sided.
Among the 4,221 EAGLE subjects, 1,946 lung cancer cases and 2,116 controls had data on family history of cancer, family history of smoking, and family history of non-malignant lung diseases and were included in the analyses (Table 1). About 75% of subjects were males. As expected, cases had lower education level than controls, while lung cancer cases had higher smoking rate than controls, e.g., higher intensity (packs per day), longer duration (years), larger total exposure (packs*years) and included more current smokers at study initiation.
Table 2 shows results of the association of family history of lung cancer, family history of smoking-related cancers and family history of any malignant tumor with lung cancer risk. There was a significant association of lung cancer risk with history of lung cancer in mother, father and siblings. Similarly, family history of smoking-related cancer in siblings and any family member was suggestively associated with increased lung cancer risk. In contrast, family history of “other tumors” was not associated with lung cancer risk. This suggests that the increased lung cancer risk we observed for any malignant tumors in any family members is likely due to family history of lung cancer and family history of smoking-related tumors.
Based on the age distribution of subjects (Supplemental Table 3), we used a cut-off of 55 years to explore the age effect on the association with family history of lung cancer in any family member. We found a higher risk in subjects < 55 years old (n=206, OR=3.26, 95% CI = 1.55–6.85) than in the older cases (n=1,740, OR=1.43, 95% CI = 1.12–1.83). We also conducted analyses stratified by gender, and found no substantial differences (data not shown).
Table 3 shows the results of the association between family history of lung cancer, family history of smoking-related tumors, and family history of any malignant tumor with lung cancer risk overall and with lung adenocarcinoma risk by smoking status (ever/never smokers). We selected adenocarcinoma because it was the only histology group with sufficient numbers of never smokers for the analysis. Sample size permitted analyses only for the subjects with cancer history in any family member or in siblings. The associations were generally stronger in never smoking subjects than ever smoking subjects, particularly for family history in siblings. Results with or without adjustment for environmental smoking exposure (passive smoking) were not substantially different among never smokers. We also conducted an analysis stratified by family history of smoking (Supplemental Table 2), which did not show major differences across the groups. However, the results were not conclusive because of the small number of subjects in the cells.
We explored the association between family history of non-malignant lung diseases, including chronic bronchitis, emphysema, pneumonia, and tuberculosis, with lung cancer risk (Table 4). Only 10–18% of relatives with family history of lung cancer also had family history of non-malignant lung diseases. Family history of chronic bronchitis was associated with increased lung cancer risk after adjustment for personal smoking exposure. In contrast, family history of pneumonia was associated with a decreased lung cancer risk. These associations were not substantially modified after further adjustment for family history of lung cancer or stratification by personal smoking status or family history of smoking (data not shown).
When stratified by age below or above 55 years old, the association between family history of chronic bronchitis in any family member and lung cancer risk was slightly stronger in the younger subjects (OR=1.76, 95% CI = 1.003–3.08) than in the older cases (OR=1.48, 95% CI = 1.21–1.81). A similar trend was observed for family history of emphysema (OR=1.34, 95% CI = 0.61–2.96 for age<55 years, and OR=1.18, 95% CI = 0.94–1.51 for age≥55 years). In contrast, the protective association with family history of pneumonia was suggestively stronger among older subjects (OR=0.78, 95% CI = 0.45–1.34 for age<55 years; OR=0.71, 95% CI = 0.59–0.87 for age≥55 years).
When stratified by histology, family history of lung cancer was associated with significantly increased risk of adenocarcinoma (OR=1.68, 95% CI = 1.28–2.20) and squamous cell carcinoma (OR=1.79, 95% CI = 1.25–2.55) but not with small cell carcinoma (OR=0.90, 95% CI = 0.51–1.59) (Table 5). A heterogeneity test showed a marginally significant difference across strata (P=0.07).
Among the non-malignant lung diseases, we found a significant association between family history of pneumonia and risk of lung cancer in the squamous cell carcinoma and adenocarcinoma patients, and an association with family history of emphysema only in the adenocarcinoma subgroup (Table 5). We found no association with small cell carcinoma, possibly because of the small number of subjects in this group; in fact the heterogeneity test showed no significant differences across the histologies. Family history of chronic bronchitis was associated with increased risk of all three histology subtypes of lung cancer. These latter analyses were not adjusted for family history of lung cancer since they did not affect the associations between family history of non-malignant lung diseases and lung cancer risk overall.
In this large case-control study with 1,946 lung cancer cases, 2,116 controls, and their first degree relatives, we found that mother’s, father’s and siblings’ history of lung cancer were significantly associated with increased lung cancer risk after adjusting for the matching variables, education, and, most notably, personal history of smoking. The associations were generally stronger among non-smokers than smokers and for the non small cell lung cancer subtype. Moreover, we observed that family history of chronic bronchitis was associated with increased lung cancer risk, and family history of pneumonia was associated with decreased lung cancer risk. These associations appeared to be independent from family history of lung cancer or family history of smoking, and varied according to the age of the lung cancer cases, with a stronger familial risk among younger cases.
Among 31 independent case-control studies published since the first report of familial aggregation of lung cancer in 1963 29, 30 studies showed consistently positive associations, with ORs ranging from 1.2 to 5.3 3, 4, 14, 15, 17, 18, 22, 23, 26–29, 34–51. Only one small study with 185 subjects showed a reverse association 52. Similarly, all 6 cohort studies showed significant positive associations 5, 32, 53–56. Therefore, our results related to family history of lung cancer are consistent with previous reports.
However, among these 31 case-control and 6 cohort studies, only 5 case-control studies included family members’ smoking information 22, 26–29. We had data on smoking history in each first degree relative. Family history of smoking can affect the risk of lung cancer via two pathways: either exposing the family members to passive smoking or affecting smoking behavior of family members (or both). We adjusted the analysis of the association between family history of lung cancer and lung cancer risk in never smokers only by passive smoking exposure, and not by family history of smoking, to take into account the first of the two potential pathways. Passive smoking exposure did not substantially affect the results in our analyses. To disentangle the effect of shared smoking exposures/habits in the families from the genetic component of family history of lung cancer, we stratified the analysis of family history of lung cancer and lung cancer risk by family history of smoking in each first-degree relative (Supplemental Table 2). The numbers were too small to allow a definitive conclusion. However, in the stratum with sufficient numbers (corresponding to family history of lung cancer in mothers stratified by FH of smoking in mothers) there was no major difference in lung cancer risk between the two strata of family history of smoking. This suggests that family history of smoking is not a major player in this association. This is in agreement with a previous smaller study in never smoker women22 and with simulation data of familial aggregation of smoking and lung cancer24. Larger studies or pooled studies are needed to conclusively address this issue.
Increased risk of squamous cell carcinoma has been consistently reported in the association with family history of lung cancer 30–32. In our analyses, family history of lung cancer was associated with elevated risk of squamous cell carcinoma and adenocarcinoma but not small cell lung cancer. These results may be due to the smaller number of subjects in the latter group or suggest that different pathways or risk factors may be involved in the etiology of small cell carcinoma vs. non small cell carcinoma. Similarly, different pathways may be involved in the etiology of lung cancer in smokers vs. non-smokers 21, as suggested by studies on gene expression 57–60, or mutations of the epidermal growth factor receptor gene 61. When we stratified the analyses by smoking status, the risk associated with family history of lung cancer was generally stronger, although not significantly, in never smokers. We observed a stronger association between family history of lung cancer and lung cancer risk among younger subjects in accordance with previous reports 7, 14, 17, 22, 23, 27, 35, 41, 62, suggesting a stronger genetic component for these cancers. Notably, family history of “other tumors” was not associated with lung cancer risk, as in previous studies 34, 55, but not in others just reporting family history of colon cancer 53, 56. This suggests that at least some of the heritable components of lung cancer risk, possibly in combination with smoking exposure, are tissue-specific. This also suggests that differential misclassification is unlikely to explain the association of family history with lung cancer risk.
Like lung cancer, chronic obstructive pulmonary disease (COPD), including chronic bronchitis and emphysema, has been linked to smoking 63. COPD patients are at high risk of developing lung cancer and recurrence from lung cancer 64, 65. We observed that both family histories of chronic bronchitis and, in younger cases, family history of emphysema were associated with increased lung cancer risk, as previously shown 45, 66. When this analysis was stratified by personal or family history of smoking status, it did not reveal major differences between smokers and never smokers. This suggests that smoking is not a strong modifier of this association. Similarly, adjustment for family history of lung cancer did not substantially change the association, and only a small percentage of relatives shared both lung cancer and non-malignant lung diseases. This accords with our understanding that the pathways that lead to COPD and lung cancer have independent and common elements that include both environmental (e.g., smoking) and inherited factors. While the epidemiological data indicate that COPD and lung cancer patients are usually smokers, the pattern of risk associated with family history suggests that there is at least some common genetic component, not necessarily related to smoking. This is in agreement with previous studies showing family history of bronchitis and emphysema in association with lung cancer risk 45, 66, and functional analyses showing impaired lung function in both relatives of lung cancer and relatives of COPD 67.
In contrast with a previous study on relatives of lung cancer cases and controls 66, we observed a reduced risk of lung cancer in association with family history of pneumonia, particularly in the older subjects. Neither personal nor familial history of smoking modified this association. Our study included a much larger number of relatives with pneumonia and specific information for each first-degree relative, which may partly explain the discrepancy between studies. Moreover, in our population, we observed a strong decrease of lung cancer risk in subjects with personal history of pneumonia (unpublished data). Although we cannot exclude that chance played a role, factors other than genes or smoking may be involved. For example, an immune system that can strongly react against malignant cells due to over-stimulation by infectious agents may partly explain these results. Further studies are necessary to confirm these findings and explore the possible mechanisms responsible for these associations.
Our study has several strengths. We took advantage of a large investigation with detailed epidemiological and clinical variables, high participation rates and in depth information on family history of cancer, smoking, and non-malignant lung diseases in first degree relatives. Moreover, the families included numerous siblings and this conferred substantial power for the analyses.
A limitation of our study is the case-control design. Without verification from medical records, we cannot exclude recall bias related to self-reported family history of cancer and non-malignant lung diseases. However, previous prospective studies have identified associations with family history of lung cancer similar to those we found. This suggests that it is improbable that recall bias substantially influenced our results. We have noted that the limited number of small cell carcinoma cases precludes a definitive answer on the association with family history of lung cancer and non-malignant lung diseases in this group. Last, the proportion of missing data due to unknown status of family history of lung cancer or non-malignant-lung diseases was relatively high, due to our conservative approach of eliminating cases if any of their first-degree relatives had unknown status. However, as shown in Supplemental Table 1, results were not substantially different when we classified the unknown group as having negative family history. Thus the missing information on the family history status in these subjects does not appear to substantially affect the results.
In conclusion, our large case-control study identified a significant increase in the risk of lung cancer, particularly of the non small cell carcinoma type, among individuals with a family history of lung cancer, regardless of family history of smoking. Notably, we observed an elevated risk of lung cancer among individuals with a family history of chronic bronchitis, and decreased risk among individuals with a family history of pneumonia, which were independent of smoking exposure and family history of lung cancer. Further studies are warranted to confirm these results and further explore the role of family history of smoking in lung carcinogenesis.
This study was supported by the Intramural Research Program of National Institutes of Health, National Cancer Institute, Division of Cancer Epidemiology and Genetics. The funding unit had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
We thank all the subjects who participated in the study and all the collaborators who contributed to its success (Collaborators are listed on the EAGLE website: http://dceg.cancer.gov/eagle)
Novelty and impact statements: Family history (FH) of lung cancer and FH of non-malignant lung diseases affected lung cancer risk independently, and did not appear to be modified by FH of smoking or personal smoking. Future epidemiological studies of lung cancer may benefit from the assessment of family history of non-malignant lung diseases, in particular chronic bronchitis and pneumonia, to obtain a more comprehensive view of the etiology of lung cancer.
Conflict of interest
We declare that we have no conflict of interest.