Cumulative Effect of Multiple Loci on Genetic Susceptibility to Familial Lung Cancer

Lung cancer is the leading cause of cancer death for both men and women in the United States. In 2008, there was an estimated 215,000 cases of lung cancer diagnosed and only 15% of those patients are expected to survive 5 years (1). Although cigarette smoking is the major risk factor for lung cancer, genetic factors also affect lung cancer susceptibility (2). For example, increased lung cancer rates are observed in several genetic syndromes including Li-Fraumeni syndrome, hereditary retinoblastoma, familial breast cancer, and Bloom syndrome (3-7). Direct evidence for a genetic predisposition to lung cancer is highlighted by several genome-wide association studies (GWAS) that have been recently done on lung cancer populations with European ancestry. These GWAS identified the associations of common variants on chromosome 15q24-25.1, 5p15.33, and 6p21.33 with lung cancer susceptibility (8-13).

Lung cancer can occur in multiple members of the same family and constitute familial lung cancer. We have previously identified the association of the chromosomal region 15q24-25.1 with familial lung cancer through a GWAS of 194 case patients with family history and 219 cancer-free control subjects collected by the Genetic Epidemiology of Lung Cancer Consortium (GELCC). The odds ratio (OR) associated with the 15q24-25.1 locus was higher than that observed in sporadic lung cancer samples (10). In addition, recent fine mapping studies identified a major candidate gene, RGS17, for the familial lung cancer susceptibility locus on chromosome 6q23-25. Two single nucleotide polymorphisms (SNP; rs9479510 and rs4083914) in the RGS17 gene are associated with familial lung cancer, but do not associate with sporadic lung cancer (14). RGS17 encodes a recently identified member of the regulator of G protein signaling (RGS) family. RGS proteins negatively regulate G protein–related signaling at least in part by accelerating the GTPase activity of Gα subunits. RGS17 is highly expressed in tumor tissues. The loss of RGS17 transcript inhibits the growth of xenografted tumors and the proliferation of tumor cells, whereas overexpression of RGS17 increases the rate of proliferation of tumor cells (14, 15).

Here, we describe the association of 5p15.33 and 6p21.33 with familial lung cancer. Taking all of the four chromosomal regions together, we assess their cumulative association with familial lung cancer and estimate the population attributable risk (PAR%) of these genetic variants underlying lung cancer susceptibility. We showed that a combination of SNPs from the chromosomal regions have a stronger association with familial lung cancer than any individual SNPs.

To confirm the association of 5p15.33 and 6p21.33 with familial lung cancer, we analyzed SNPs within these two chromosomal regions from our previous study (10). We have previously conducted a GWAS among individuals with a familial history of lung cancer using the Affymetrix SNP chips. These individuals are members of families with three or more members with lung cancer that were collected as part of the GELCC collections. To ensure genetic independence among subjects, each case patient with familial lung cancer was chosen from one high-risk lung cancer family. All the case patients in this study are histologically confirmed non–small cell lung cancer. Noncancer control subjects were obtained from a combination of unaffected spouses from GELCC families and of unaffected individuals from the Coriell Institute for Medical Research and the Fernald Medical Monitoring Program. These control subjects had no blood relationship with any selected case patients. Totally, 194 case patients with familial lung cancer and 219 cancer-free control subjects were used. Detailed information on the GELCC samples was described in Supplementary Table S1.

We analyzed 144 SNPs on 5p15.33 (1.1-1.7 Mb) and 2,337 SNPs on 6p21.33 (27.0-34.0 Mb) from our Affymetrix SNP chips. To comprehensively cover these two regions, we typed additional 11 SNPs on 5p15.33 and 21 SNPs on 6p21.33 in the GELCC collection. These additional SNPs are those showing strong association in previous studies (11, 13) but were not presented in the Affymetrix SNP chips. SNP genotyping was done using the Sequenom mass array spectrometry system at the Human Genetics Division Genotyping Core of Washington University (St. Louis, MO). Genotyping results are initially evaluated according to the fraction of assay successes per plate. The genotyping call rate is >90% for each SNP in our study. In addition to assay success rates, genotyping plates are reviewed for results from positive and negative DNA control wells that are organized in specific patterns to assist in the quality check (QC) process and to ensure correct plate orientations during processing and data review.

Hardy-Weinberg equilibrium for each SNP was examined among control subjects with the use of Fisher exact test. The statistical significance of the association between SNP allele and disease status was assessed primarily with the Cochran-Armitage trend test with a 1 degree of freedom and with the Fisher exact test, implemented in the PLINK software.¹¹

We tested the cumulative effects of the four chromosomal regions (5p15.33, 6p21.33, 15q24-25.1, and 6q23-25/RGS17) on familial lung cancer by counting the number of risk allele at each locus in logistic regression model. One SNP was chosen from each of the four chromosomal regions. The chosen SNPs are those that have been validated with association with sporadic lung cancer and/or familial lung cancer. To improve the stability of estimates of OR, we combined the sporadic lung cancer samples from three recent GWAS (8, 9, 13) with our familial samples in the data analysis. The cumulative ORs for subjects carrying different copies of risk alleles were estimated by comparing them with those carrying none of these risk alleles.

PAR% was estimated for each locus, which defines what percentage of the total risk for lung cancer is due to the genetic effect of that locus: PAR = Σp_i × (OR_i − 1)/(Σp_i × (OR_i − 1) + 1), in which p_i is the prevalence of risk allele at ith locus associated with lung cancer among control subjects, and OR_i is OR of risk allele at ith locus. The joint PAR was calculated on the basis of the individual PAR of each associated SNP: 1 − ∏(1 − PAR_i) in which PAR_i is the individual PAR for each associated SNP.

In the GELCC collection of 194 familial cases and 214 cancer-free controls, we observed a statistically significant association between 5p15.33 and familial lung cancer (Supplementary Table S2; Table 1). Twelve SNPs on the 5p15.33 have P value of <0.01. Four of five SNPs identified in previous studies (11, 13) were confirmed in the GELCC collections. Among these SNPs, rs31489 is the most significant one with P value of 2.0 × 10⁻⁴ (OR, 0.57; 95% CI, 0.42-0.77). The frequency of the minor allele A of rs31489 was 32.49% in cancer patients versus 45.79% in disease-free subjects. Subjects carrying one copy of minor allele A are at 1.75-fold reduced risk of developing lung cancer. The association of the 5p15.33 locus with familial lung cancer remains statistically significant after adjusting sex, age, and smoking quantity (P = 8.0 × 10⁻⁴). We did not observe the association of 5p15.33 with smoking quantity in the familial samples (Supplementary Table S3).

We then compared the association of 5p15.33 in cases from families with three affected members and in cases from families with four or more affected members. Generally, the association of 5p15.33 tends to be stronger in cases from families with four or more affected members than in cases from families with three affected members. However, there is no statistically significant difference in allelic ORs between these two groups of familial cases (Table 2). When comparing the association of 5p15.33 in familial and sporadic cases, we observed that the effect size of 5p15.33 is significantly larger in familial case than in sporadic cases (P = 4.8 × 10⁻³ from the Woof's test; OR, 0.57; 95% CI, 0.42-0.77 in familial cases; OR, 0.88; 95% CI, 0.84-0.84 in sporadic cases; Table 3; ref. 13).

In the GELCC collections, the susceptibility region on 5p15.33 contains three genes: TERT, CLPTM1L, and SLC6A3. SNPs with the strongest associations are located within CLPTM1L (Fig. 1A). Interestingly, we identified an additional significant SNP rs466630 (P = 4.5 × 10⁻³; OR, 1.85; 95% CI, 1.24-2.77) that is independent of previously reported SNPs (r² < 0.001; Supplementary Table S2). This SNP is located within SLC6A3.

We observed a weak association between 6p21.33 and familial lung cancer in the GELCC collections (Supplementary Table S4; Table 1). Among the three SNPs on 6p21.33 reported in a recent study (13), rs3117582 achieved a P value of 0.09 (OR, 1.47; 95% CI, 0.94-2.31). The minor allele C of rs3117582 is enriched in cases compared with disease-free controls, which is consistent with that observed in the previous study (13). The weak association observed on 6p21.33 may be due to the small sample size of the GELCC collections in this study and to the low population frequency of the risk allele of rs3117582. However, we found that several SNPs (e.g., P = 6 × 10⁻⁴; OR, 1.71; 95% CI, 1.25-2.33 for rs1634718) nearby rs3117582 show stronger associations with familial lung cancer (Supplementary Table S4; Fig. 1B). There is no significant difference in the strength of association of 6p21.33 by the number of familial affecteds in the kindred of origin (Table 2).

We then evaluated what percentage of the total risk for familial lung cancer is due to the genetic effect of the 5p15.33, 6p21.33, and 15q24-25.1. The 15q24-25.1 region has been recently reported to be associated with both familial and sporadic lung cancer (8-10, 12). At least two common variants on 15q24-25.1, SNPs rs8034191and rs1051730, were confirmed in the GELCC collections (Supplementary Table S5 and S6). We observed that 5p15.33, 6p21.33, and 15q24-25.1 account for 7.5%, 2.7%, and 8.5% risk of lung cancer in the familial population, respectively. The three genetic regions contribute to a total of 17.6% PAR for familial lung cancer (Supplementary Table S7; Table 4). The risk of lung cancer ranged from 2- to 4-fold (that is cumulative OR, 2-4) among those subjects who had at least one copy of risk allele at each of the three regions compared with subjects without any of risk alleles.

RGS17 has been recently identified as a major candidate gene for the familial lung cancer susceptibility locus on chromosome 6q23-25 (Supplementary Table S5 and S6; ref. 14). We therefore took RGS17 into consideration in the analysis of the cumulative association of familial lung cancer. The 6q23-35/RGS17 accounts for an additional 20.66% PAR for lung cancer in the familial samples. The total risk of familial lung cancer contributed by these four genetic regions was further increased to 34.6% (Table 4). Analysis of a combination of SNPs from these regions provided a stronger cumulative association with familial lung cancer (P = 6.70 × 10⁻⁶). The cumulative OR associated with these four regions ranged from 3- to 11-fold among those subjects who had at least one copy of risk allele at each region, compared with subjects without any of the risk factors. However, <2% carriers of risk alleles have cumulative ORs larger than five in the populations (Table 5).

In this study, we replicated the overall association of 5p15.33 and 6p21.33 with familial lung cancer. We found that the SNP rs31489 shows the strongest evidence of familial lung cancer association on 5p15.33 (P = 2 × 10⁻⁴; OR, 0.57; 95% CI, 0.42-0.77), whereas rs3117582 shows a relatively weak association on 6p21.33 (P = 0.09; OR, 1.47; 95% CI, 0.94-2.31). Although each of these chromosomal regions was only moderately associated with lung cancer, we observed that they have a strong cumulative association with the disease. We show 3- to 11-fold increase in lung cancer risk based on the cumulative association of 5p15.33, 6p21.33, 15q24-25.1, and 6q23-25/RGS17 among those subjects who had at least one copy of risk allele at each of the four regions compared with subjects without any of the risk factors. These four genetic regions account for 34.6% of familial lung cancer. These results further confirm the importance of these genomic regions underlying lung cancer susceptibility.

Our results should encourage further studies characterizing candidate genes for each of the loci and the mechanism about how these genetic variants contribute to the risk of lung cancer. The candidate genes in the 5p15.33 locus include TERT and CLPTM1L in previous studies (11, 13). In the present study, fine mapping was done on the 5p15.33 locus using a more dense set of SNPs from the Affymetrix SNP 6.0 array and additional SNPs genotyped by the Sequenom mass array spectrometry system. The familial data show very strong associations on the 5p15.33, and the strongest association was located within CLPTM1L. Several SNPs in SLC6A3 also showed strong associations with familial lung cancer (Fig. 1A). Gain in 5p15.33 is one of the most frequent events observed in early-stage non–small cell lung cancer (16). The 5p15.33 locus has been recently identified to associate with multiple cancer types including lung, urinary bladder, prostate, and cervical cancer (17). TERT is the reverse transcriptase component of telomerase and is an attractive candidate for the 5p15.33 locus. Telomere attrition limits the replicative potential of most somatic cells. In contrast, tumor cells acquire immortality by continuous telomere maintenance, which is predominantly due to the transcriptional upregulation of TERT (18). Interestingly, TERT has also been found to directly promote proliferation (19, 20). CLPTM1L, originally named for its relationship to a gene in the cleft palate susceptibility, was founded to sensitize ovarian cancer cells to cisplatin-induced apoptosis (21). SLC6A3 (also named the dopamine transporter, DAT1) mediates the active reuptake of dopamine from the synapse and is a principal regulator of dopaminergic neurotransmission. The SLC6A3 gene has been implicated in human disorders such as Parkinsonism, Tourette syndrome, and substance abuse (22).

We also observed a weak association on 6p21.33 in the familial data. Among many genes at this locus, BAT3 and MSH5 are attractive candidates. BAT3/SCYTHE is known to regulate apoptosis and is essential for p53 acetylation and p53-mediated DNA damage response (23). Thus, variants with decreased function may confer the risk of lung cancer through decreased apoptosis in response to DNA damage. MSH5 is a member of the MUTS family of proteins involved in DNA mismatch repair and can stabilize the double-strand break repair intermediately (24). Missense mutations found in MSH5 have been shown to result in increased tolerance to killing by DNA alkylating agents (25). It is conceivable that variants of MSH5 could confer tolerance to DNA alkylation–induced apoptosis.

We have previously identified the association between 15q24-25.1 and familial lung cancer in the GELCC collections. The 15q25.1 contains IREB2, LOC123688, PSMA4, CHRNA5 CHRNA3, and CHRNB4. The CHRN genes encode for subunits of the nicotinic acetylcholine receptor subunits, seem biologically relevant, and are initially attractive candidates. Among the remaining genes in the locus, IREB2 encodes an iron regulator protein 2 and plays a central role in maintaining cellular iron homeostasis (26); PSMA4 encodes a structural protein of the 20S proteasome core (27, 28); and LOC123688 is a hypothetical gene.

Our study is unique because of its use of familial lung cancer samples from the GELCC collections. Each case was chosen from one family who has three or more members with lung cancer. However, several caveats for our findings should be acknowledged. First, familial lung cancer occurs in <1 of 1,000 lung cancer patients diagnosed. This results in small sample sizes in family studies and limits statistical power in the association analysis. Second, lung cancer is a late-onset disease. Tobacco smoking is well established as the major risk factor for lung cancer, contributing to a 10-fold increase in risk in long-term smokers compared with nonsmokers (29). To increase the statistical power to detect genetic variants and reduce confounding effects, in addition to spousal controls, we chose old, healthy, heavy smokers as controls. This sample strategy may lead to biased estimation of covariate effects and risk effects at genetic loci in the logistical regression model. However, these caveats should not affect our conclusion that a combination of SNPs from the chromosomal regions has a stronger association with familial lung cancer than any individual SNPs at chromosome 5p15.33, 6p21.33, 6q23-25, and 15q24-25.1. As a whole, the risk of lung cancer is increased significantly among those subjects who had at least one copy of risk allele at these chromosome regions.

No potential conflicts of interest were disclosed.

We thank the Fernald Medical Monitoring Program for sharing their biospecimens and data with us, the lung cancer families who participated in this research, the Vanderbilt University Microarray Shared Resource, Washington University Genotyping Core, Mayo Clinic Genotyping Shared Resource (supported in part by P30CA 15083) for the high caliber service, and Shaw Levy and Jennifer Baker for their assistance in various aspects of this work.

Grant Support: NIH grants U01CA76293 (GELCC), R01CA058554, R01CA093643, R01CA099147, R01CA099187, R01ES012063, R01ES013340, R03CA77118, R01CA80127, P30ES06096, P50CA70907 (Specialized Program of Research Excellence), N01HG65404, N01-PC35145, P30CA22453, R01CA63700, DE-FGB-95ER62060, Mayo Clinic intramural research funds, and Department of Defense VITAL grant. This study was supported in part by NIH, the Intramural Research Programs of the National Cancer Institute, and the National Human Genome Research Institute.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.