|Home | About | Journals | Submit | Contact Us | Français|
Background. Analysis of candidate genes in individual studies has had only limited success in identifying particular gene variants that are conclusively associated with lung cancer risk. In the International Lung Cancer Consortium (ILCCO), we conducted a coordinated genotyping study of 10 common variants selected because of their prior evidence of an association with lung cancer. These variants belonged to candidate genes from different cancer-related pathways including inflammation (IL1B), folate metabolism (MTHFR), regulatory function (AKAP9 and CAMKK1), cell adhesion (SEZL6) and apoptosis (FAS, FASL, TP53, TP53BP1 and BAT3). Methods. Genotype data from 15 ILCCO case–control studies were available for a total of 8431 lung cancer cases and 11072 controls of European descent and Asian ethnic groups. Unconditional logistic regression was used to model the association between each variant and lung cancer risk. Results. Only the association between a non-synonymous variant of TP53BP1 (rs560191) and lung cancer risk was significant (OR=0.91, P=0.002). This association was more striking for squamous cell carcinoma (OR=0.86, P=6×10−4). No heterogeneity by center, ethnicity, smoking status, age group or sex was observed. In order to confirm this association, we included results for this variant from a set of independent studies (9966 cases/11722 controls) and we reported similar results. When combining all these studies together, we reported an overall OR=0.93 (0.89–0.97) (P=0.001). This association was significant only for squamous cell carcinoma [OR=0.89 (0.85–0.95), P=1×10−4]. Conclusion. This study suggests that rs560191 is associated to lung cancer risk and further highlights the value of consortia in replicating or refuting published genetic associations.
It has been long recognized that there is a genetic component for lung cancer based on familial studies and analysis of family cancer history in case–control studies (1–5). However, progress in identifying specific susceptibility loci and genes has been slow, mainly due to inadequate study designs, underpowered sample sizes and preferential reporting of false-positive findings. Recently, several genome-wide association studies (GWAS) reported lung cancer susceptibility loci at 15q25, 6p21 and 5p15.33, providing additional evidence of a genetic contribution to lung cancer (6–10).
Although replication of findings is fundamental to the GWAS approach, the replication of significant variants in studies based on the more traditional candidate gene approach may also be an important procedure. To confirm the role of candidate genetic variants that were found to be associated with lung cancer risk, we established a fast-track replication mechanism for testing genetic variants within the framework of the International Lung Cancer Consortium (ILCCO). ILCCO was established in 2004 with the aim of sharing comparable data from ongoing lung cancer case–control and cohort studies. The overall objectives are to increase statistical power, especially for subgroup analyses, reduce duplication of research efforts, replicate novel findings and afford substantial cost savings through large collaborative efforts. Details on the organization of the consortium have been published previously (11).
In order to prioritize genetic variants for this rapid replication within the consortium, a procedure was established to replicate genetic variants newly associated with risk in previous lung cancer studies. Two criteria were used to prioritize the genetic variants: (i) a similar significant main effect (P<0.05) in at least two studies, both of which had sample sizes of at least 500 case–control pairs or (ii) a strongly significant main effect (P<0.001) from a single study. Ten potential lung cancer susceptibility variants meeting the criteria were proposed by the consortium members for replication (Table I). These variants belonged to genes from various cancer-related pathways including inflammation (IL1B), folate metabolism (MTHFR), regulatory function (AKAP9 and CAMKK1), cell adhesion (SEZL6) and apoptosis (FAS, FASL, TP53, TP53BP1 and BAT3). The description of the selected variants is detailed in supplementary Table I (available at Carcinogenesis Online).
In the current study, we conducted a coordinated genotyping of 10 potential lung cancer susceptibility variants in 15 studies. From these studies, six were conducted in the USA, six in Asia and three in Europe. Study designs are briefly outlined in Table II, and more detailed information for some of these studies has been published previously (6,20–31). Studies are referred to by study location or coordinating institute.
Eligibility criteria based on age were applied in two studies: ≤50 years for the German study and <65 years for the University of California, Los Angeles study. Eligibility criteria based on smoking status were applied in three studies: the Singapore study was restricted to never-smoking women, whereas the MD Anderson and Norway studies included only ever smokers. In all studies, cases were histologically or cytologically confirmed, except in the NCI-China study where the inclusion criteria comprise a positive histology or cytology, or clinically diagnosed cases that died within a 1 year period. Therefore, the NCI-China study did not have information on histology for all cases. The Norway study was restricted to non-small cell carcinomas cases. The control group in most of the studies was frequency matched to the cases on age and sex, whereas some also matched on ethnicity, residence area or smoking status, and two studies did not apply any matching factors (Kyushu University and Mayo clinic studies).
Written informed consent was obtained from all study subjects, and approval from the relevant ethics board was obtained at each study center. Genotype data were available for a total potential number of 8705 cases and 11562 controls, of whom 68% where European descent, 28% were Asian and 4% were of different ethnic groups (African-American, Hispanic, Native Hawaiian and American Indian). Only European descent (5876 cases and 7874 controls) and Asian (2555 cases and 3198 controls) ethnic groups were included in the present study. The remaining ethnic groups were excluded due to small sample size.
Genomic DNA was extracted from blood samples in all centers, except in the Penn State College of Medicine and the NCI-China studies for which DNA was extracted from oral buccal mucosa cells. Several studies had independently genotyped some of the selected variants prior to this initiative using their own protocol and did not genotype additional variant (German, MD Anderson, NCI-China, Aichi and Nanjing-China studies). The genotyping procedures of these studies are described elsewhere (6,23,24,28,31). For all other centers, the genotyping was done locally using Taqman (Applied Biosystems, Forster City, CA). Three studies (Kyushu, Seoul and Singapore studies) genotyped a subset of the selected single nucleotide polymorphisms (SNPs) due to DNA availability. Procedures for inter-laboratory quality control were applied for the German study and all centers that used Taqman probes: each genotyped a generic series of common DNAs (either SNP500, HapMap CEU trios or International Agency for Research on Cancer generic series) using their local genotyping facility. The genotype concordance across studies was subsequently computed for each genotyping assay. Discrepancies rates for all assays were <5%. Studies with a call rate of <90% for a variant were excluded from analysis for that variant. Table III shows the number of cases and controls included in analysis and minor allele frequency among controls for each variant and each study by ethnic group. It should be noted that in this replication study, centers that have generated the hypothesis on a variant were excluded from analysis for the corresponding variant. As the number of participating studies differed for each variant, subjects included in analysis varied greatly from a variant to another especially in Asians.
We used a Chi-squared test with one degree of freedom to verify that allele distributions for each of the 10 SNPs were in Hardy–Weinberg equilibrium within each study and among European descent controls and Asian controls separately. A Bonferroni correction for multiple tests was applied for the Hardy–Weinberg equilibrium test giving an indicative P-value of 5×10−4 (based on ~100 tests carried out). All genotype distributions respected Hardy–Weinberg equilibrium.
Smoking status was categorized into three categories as never, former and current smokers. Former smokers were defined as smokers who stopped smoking for at least 2 years prior to the diagnosis (for cases) or interview (for controls). In the Norway and NCI-China studies, it was not possible to distinguish former from current smokers and ever smokers were considered as current smokers.
We used unconditional logistic regression to estimate odds ratios (ORs) and 95% confidence intervals. Pooled ORs were calculated using individual-level data from ILCCO studies. The heterozygous and homozygous carriers of the minor allele were each compared with the homozygous carriers of the major allele. OR per allele or P-values for trend were calculated assuming a log-additive genetic model with one degree of freedom. European descent and Asian subjects were analyzed as separate groups as well as together. Models were adjusted for study-matching variables and potential confounders including age, sex, smoking status and center when appropriate. We conducted stratified analysis by smoking status and age of diagnosis to evaluate effect modification. We also analyzed the association of genetic variants and lung cancer risk by major histological subtypes (squamous cell carcinoma, adenocarcinoma and small cell carcinoma). Heterogeneity of ORs across the studies and across the stratification groups (age, sex and ethnicity) was assessed using the Cochran's Q-test. All analysis was conducted with SAS 9.1.
In order to confirm the association we found between rs560191 (TP53BP1) and lung cancer risk in the ILCCO sample, we have genotyped additionally 1996 cases and 3487 controls from two further case–control studies (Poland and EPIC-lung) using Taqman. We have also incorporated results from several GWAS on lung cancer (National Cancer Institute, Texas, Canada, Tromso, CARET) including a total of 7970 cases and 8235 controls, as this variant is tagged on the Illumina panel by the proxy rs2602141 (R2=1.00, D′=1.00).
The Polish study is a hospital-based case–control study conducted in Szczecin between 2004 and 2007 (32). No data on smoking status were available for this study. EPIC-lung study is a case–control study nested in the EPIC cohort (European Prospective Investigation into Cancer and Nutrition) that was conducted in 10 European countries (Sweden, Denmark, The Netherlands, UK, France, Germany, Spain, Italy, Greece and Norway). This study was described in detail previously (7,33).
The NCI, Texas, Canada, HUNT2/Tromso and CARET GWAS studies were part of a meta-analysis that was conducted previously (34). Briefly, the NCI GWAS (National Cancer Institute) was drawn from four different studies: the Environment and Genetics in Lung Cancer Etiology, a population-based case–control study conducted in Italy in 2002–2005 (35); the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study (36), a cohort study including male smokers enrolled in Finland in 1985–1993; the Prostate, Lung, Colon, Ovary Screening Trial, a randomized trial conducted in 10 US centers between 1992 and 2001 (37) and the Cancer Prevention Study II Nutrition Cohort, a cohort study conducted across all US in 1992–2001(38). The Texas study is a hospital-based case–control study including only smokers; cases with non-small cell lung cancer were diagnosed at the University of Texas MD Anderson Cancer Center since 1991 (6). The CARET cohort study (Carotene and Retinol Efficacy Trial) was conducted in six US centers between 1983 and 1994 and included only smokers with a smoking history of at least 20 pack-years (39). The HUNT2/Tromso study included cases and controls from two large population-based studies: the North Trondelag Health Study (HUNT2) conducted in 1995–1997 and the Tromso IV study conducted in 1994–1995. The Canada study is a hospital-based case–control study conducted at the University of Toronto and the Samuel Lunenfeld Research Institute in 1997–2002 (7).
All these cases and controls were of European descents. As no heterogeneity was detected across studies, we used a fixed-effects model to calculate ORs and 95% confidence intervals for all studies combined.
Table IV shows the ORs of lung cancer for the 10 selected variants, for all subjects and stratified by ethnicity. No significant heterogeneity across ethnic groups was reported. We did not observe an association between any of the variants and risk of lung cancer, except rs560191 (TP53BP1), which was significantly associated with lung cancer risk in European descents and Asians [overall OR per rare allele=0.91 (0.86–0.97), P-value=0.002]. When considering both ethnic groups, no significant heterogeneity by study, histology, smoking status, age group or sex was observed (supplementary Figure 1 is available at Carcinogenesis Online). However, even if the heterogeneity by histology was non-significant (P=0.15), the association was more striking for squamous cell carcinoma [OR=0.86 (0.79–0.94), P=6×10−4] than for adenocarcinoma [OR=0.97 (0.89–1.04), P=0.39] or small cell carcinoma [OR=0.91 (0.80–1.03), P=0.14].
In order to confirm the findings on rs560191, we have genotyped this variant in two additional studies (EPIC-Lung and Poland) and we have also incorporated results from several GWAS studies on rs2602141 (a proxy of rs560191). The results of this meta-analysis are reported in Table V. The associations found are similar to those of the ILCCO study. Overall, we reported an OR of 0.95 (0.91–0.99) (P=0.02). The association was significant for squamous cell carcinoma [OR=0.92 (0.86–0.99), P=0.03] but not for adenocarcinoma [OR=0.95 (0.89–1.02)] and small cell carcinoma [OR=0.98 (0.88–1.09)] (P-heterogeneity by histology=0.60).
The result of the combined analysis of the ILCCO study and the additional studies is shown in Figure 1. We reported an overall OR of 0.93 (0.89–0.97) (P=0.001), and we found that this association concerned mostly squamous cell carcinoma [OR=0.89 (0.85–0.95), P=1×10−4]. When combined with the initial association that generated the hypothesis reported by Rudd et al. [OR=0.85 (0.77–0.92)], P=9×10−4, Table II), the allelic OR was 0.91 (0.88–0.95), P=6×10−6.
Stratified analyses were also performed for the other variants considered, and the results showed no consistent patterns (results not shown).
This replication study was based on common variants that were previously found associated with lung cancer risk in a candidate gene approach. We evaluated the effect of 10 variants proposed by the ILCCO members on lung cancer risk in a pooled analysis of 15 studies. The large sample size collected in European descent and Asian groups permitted us to analyze potential effect modification in subgroups of interest (by smoking status, age of onset and histological subtypes). We found that rs560191 (TP53BP1) was associated with a decreased risk of lung cancer. This association was more striking for squamous cell carcinoma. This result was replicated in a meta-analysis of eight independent studies based on an additional 9966 cases and 11722 controls.
We did not find an association between lung cancer risk and any of the other variants. With the exception of MTHFR C677T, none of these variants had been previously analyzed in the context of a pooled or a meta-analysis of lung cancer studies (40,41).
To our knowledge, only one study has investigated the role of TP53BP1 variants in lung cancer susceptibility (12). This study analyzed the association between lung cancer and 1476 non-synonymous variants from 871 candidate cancer genes in a large case–control study conducted in UK (1529 cases and 2797 controls) and reported a decreased risk of lung cancer among rare allele carriers of rs560191 [OR per allele=0.85 (0.77–0.93)]. This result was replicated in the present study [OR=0.93 (0.89–0.97), P=0.001]. This association remains significant after a Bonferroni correction for multiple tests based on 10 independent tests carried out (corresponding to the number of variants analyzed) and then considering a threshold P-value of 0.005. No heterogeneity across studies and ethnic groups was observed in the overall analysis. However, even if the test for heterogeneity was non-significant across histology groups (P=0.23), we reported a stronger association between squamous cell carcinoma and rs560191 [OR =0.89 (0.85–0.95), P=1×10−4].
rs560191 is a non-synonymous variant of TP53BP1, although no report on any functional relevance of this variant was published so far. TP53BP1 is one of the TP53-regulated genes and it is known to be involved in DNA damage–signaling pathways, in checkpoint signaling and in DNA repair activities (42,43). It is also known to interact with the DNA-binding domain of TP53 to enhance TP53-mediated transcriptional activation (44). One study suggested that TP53BP1 may have protective effects on squamous cell carcinoma of the head and neck risk but such effects were confined to TP53 variant allele/haplotype carriers (45). In the present pooled analysis, only one study had data for both TP53BP1 and TP53 intron 3, and we did not observe any interaction between these two variants in lung cancer risk (data not shown).
Several case–control studies have reported previously on the association between the MTHFR C677T variant and lung cancer risk, including predominantly European (18,46–48) and Asian studies (16,17,49,50). Some of these studies (16,18,28,46) also analyzed possible effect modification by folate intake, as the MTHFR enzyme is involved in folate metabolism. A recent meta-analysis of these studies (41) suggested no evidence for a major role of the MTHFR C677T polymorphisms in carcinogenesis of lung cancer. Another recent meta-analysis (40), including stratified analysis by folate intake levels, observed a non-significant increased risk of lung cancer among MTHFR 677 TT individuals with low folate intake. In the present study, no association between MTHFR C677T and lung cancer risk was observed; however, information on folate intake was not available.
Few studies have investigated the association between the other variants considered and lung cancer risk. AKAP9 M463I, CAMKK1 E375G and BAT3 S625P were detected as candidate variants in the same study mentioned above that highlighted TP53BP1 D353E (12). Interestingly, a variant in BAT3 was recently found strongly associated with lung cancer risk (10), although D′=1.0 and r2<0.08 with BAT3 S625P. Association between IL1B C3954T and lung cancer risk was reported in a study from US that evaluated a panel of 59 variants in 37 inflammation-related genes (13). Another study also found such an association with this variant, but only among former smokers and men (51). Conversely, a study conducted in Norway found no association between IL1B C3954T and lung cancer risk but reported significant associations with other variants in this gene (52,53). SEZ6L M430I was detected as a candidate variant affecting lung cancer risk in a pilot study testing 83715 SNPs and this association was confirmed in two replication studies (14). FAS G1377A and FASL C844T were investigated in a case–control study conducted in China (15). This study reported an increase risk of lung cancer among wild allele carriers for both SNPs and a multiplicative gene–gene interaction. These results were not replicated in two other independent studies, one conducted in Korea (54) and the other conducted in US (51). Association between the TP53 intron 3 polymorphism and lung cancer risk was examined by three studies. Two studies conducted, respectively, in US and Central Europe (19,20) reported that carrying 16 bp duplications in TP53 intron 3 increased the risk of lung cancer, whereas another study conducted in Sweden (55) did not detect any association.
The present study shows the importance of consortia in replicating potential significant results from candidate gene studies. Only for 1 of 10 variants was the association replicated, suggesting that initial positive results for the other SNPs may be false.
In conclusion, among the 10 variants selected for replication based on prior evidence of a potential association with lung cancer risk, we reported significant results only for rs560191 (TP53BP1). Rare allele carriers of this variant were found to have a modest decreased risk of lung cancer. This association concerned mainly squamous cell carcinomas. Subsequent studies from ILCCO will focus on replication of lung cancer susceptibility variants identified by GWAs.
US National Institutes of Health, National Cancer Institute (R03 CA133939-01); Central Europe study—World Cancer Research Fund and European Commission's INCO-COPERNICUS Program (IC15-CT98-0332); Norvegian study—The Norwegian Research Council; Norwegian Cancer Society; Aichi Cancer Center study—Grants-in-Aid for Scientific Research, Ministry of Education, Science, Sports, Culture and Technology of Japan; MD Anderson study—National Institutes of Health (CA127219, CA55769, CA121197); Penn State study—Public Health Service grants (K99 CA131477), National Institutes of Health (P01 CA68384, K07 CA104231); NCI-China study—Intramural National Cancer Institute program (N01 CO12400); Singapore study—Singapore National Medical Research Council; Seoul study—Eco-technopia 21 project, Ministry of Environment, Republic of Korea; “Deutsche Krebshilfe” (70-2387, 70-2919); Mayo Clinic study—National Institutes of Health (CA77118, CA80127, CA84354); Norris Cotton Cancer Center study—National Center for Research Resources, National Institutes of Health (P20RR018787). Environment and Genetics in Lung Cancer Etiology, Prostate, Lung, Colon, Ovary Screening Trial, Alpha-Tocopherol, Beta-Carotene Cancer Prevention studies, genotyping of Alpha-Tocopherol, Beta-Carotene Cancer Prevention study, Cancer Prevention Study II Nutrition Cohort and part of Prostate, Lung, Colon, Ovary Screening Trial—Intramural Research Program, National Institutes of Health, National Cancer Institute, Division of Cancer Epidemiology and Genetics; Alpha-Tocopherol, Beta-Carotene Cancer Prevention study—US Public Health Service contracts, National Cancer Institute (N01-CN-45165, N01-RC-45035, N01-RC-37004); PLCO study was also supported by individual contracts from the National Cancer Institute to the University of Colorado Denver (NO1-CN-25514); Georgetown University (NO1-CN-25522); Pacific Health Research Institute (NO1-CN-25515); Henry Ford Health System (NO1-CN-25512); University of Minnesota (NO1-CN-25513); Washington University (NO1-CN-25516); University of Pittsburgh (NO1-CN-25511); University of Utah (NO1-CN-25524); Marshfield Clinic Research Foundation (NO1-CN-25518); University of Alabama at Birmingham (NO1-CN-75022); Westat (NO1-CN-25476); University of California, Los Angeles (NO1-CN-25404); The Cancer Prevention Study II Nutrition Cohort was supported by the American Cancer Society; The NIH Genes, Environment and Health Initiative partly funded DNA extraction and statistical analyses (HG-06-033-NCI-01, RO1HL091172-01); genotyping at the Johns Hopkins University Center for Inherited Disease Research (U01HG004438, NIH HHSN268200782096C); study coordination at the GENEVA Coordination Center (U01 HG004446) for Environment and Genetics in Lung Cancer Etiology and part of PLCO studies. German study—BMBF, Germany (Competence Network Radiation Research, project: individual susceptibility and genomic instability); KORA research platform—the Helmholtz Center Munich, German Research Center for Environmental Health.
Conflict of Interest Statement: None declared.
Paolo Vineis (Servizio di Epidemiologia dei Tumori, Università di Torino and CPO-Piemonte, Turin, Italy; Department of Epidemiology and Public Health, Imperial College, London, UK), Francoise Clavel-Chapelon [INSERM (Institut National de la Santé et de la Recherche Médicale), ERI 20, EA 4045, and Institut Gustave Roussy, Villejuif, F-94805 France], Domenico Palli [Molecular and Nutrional Epidemiology Unit, Cancer Research and Prevention Institute (ISPO), Florence, Italy], Rosario Tumino (Cancer Registry and Histopathology Unit, Department of Oncology, “Civile M.P.Arezzo” Hospital, ASP 7, Italy), Vittorio Krogh (Nutritional Epidemiology Unit, Fondazione IRCCS Istituto Nazionale Tumori, Milan, Italy), Salvatore Panico (Dipartimento di Medicina Clinica e Sperimentale,Universita di Napoli, Federico II, Naples, Italy), Carlos A. González (Unit of Nutrition, Environment and Cancer, Cancer Epidemiology Research Programme, Catalan Institute of Oncology. L'Hospitalet de Llobregat, Barcelona, Spain), José Ramón Quirós (Jefe Sección Información Sanitaria, Consejería de Servicios Sociales, Principado de Asturias, Oviedo, Spain), Carmen Martínez (Escuela Andaluza de Salud Pública, Granada, Spain), Carmen Navarro (Epidemiology Department, Murcia Health Council, Murcia, Spain; CIBER Epidemiologia y Salud Publica (CIBERESP), Spain), Eva Ardanaz (Registro de Cáncer de Navarra, Instituto de Salud Pública, Gobierno de Navarra, Pamplona, Spain; CIBER Epidemiologia y Salud Publica (CIBERESP), Spain), Nerea Larrañaga (Subdirección de Salud Pública de Gipuzkoa, Gobierno Vasco, San Sebastian, Spain), Kay Tee Khaw (University of Cambridge, UK), Timothy Key (Cancer Research UK, University of Oxford, Oxford, UK), H.Bas Bueno-de-Mesquita [Centre for Food and Health, National Institute for Public Health and the Environment (RIVM), Bilthoven, The Netherlands], Petra.HM Peeters (Julius Center for Health Sciences and Primary Care, Department of Epidemiology, University of Utrecht, Utrecht, The Netherlands), Antonia Trichopoulou (WHO Collaborating Center, Department of Hygiene, Epidemiology and Medical Statistics, University of Athens, and Hellenic Health Foundation, Athens, Greece), Rudolf Kaaks (Division of Clinical Epidemiology, German Cancer Research Centre, Heidelberg, Germany), Heiner Boeing (Department of Epidemiology, Deutsches Institut für Ernährungsforschung, Potsdam-Rehbrücke, Germany), Göran Hallmans (Department of Public Health and Clinical Medicine, University of Umeå, Umeå, Sweden), Kim Overvad (School of Public Health, Aarhus University, 8000 Aarhus C, Denmark), Anne Tjønneland (The Danish Cancer Society, Institute of Cancer Epidemiology, Copenhagen, Denmark), Merethe Kumle (Institute of Community Medicine, University of Tromsø, Tromsø, Norway), Elio Riboli (Department of Epidemiology and Public Health, Imperial College, London, UK).