Abstract:Abstract Background: Impaired lung function (LF) is strongly associated with increased lung cancer risk. However, since airflow obstruction is a diagnostic criterion for obstructive lung disease, and a consequence of tobacco smoking, isolating the causal relationship between LF and lung cancer has remained a challenge. Methods: We investigated 3 standardized (mean=0, standard deviation=1) LF metrics: forced expiratory volume in 1 second (FEV1), forced vital capacity (FVC), and FEV1/FVC. To evaluate the causal relevance of LF in lung cancer etiology we conducted: i) survival analyses in the UK Biobank cohort (UKB); and ii) Mendelian Randomization (MR) analyses using genetic instrumental variables (IVs) developed in UKB and tested using individual-level data from the OncoArray, a genome-wide array with in-depth coverage for common cancers. Results: 702 incident lung cancers were diagnosed in 484,194 UKB participants during follow-up. Cox regression was used to estimate hazard ratios (HR) and 95% confidence intervals (CI), adjusted for age, sex, smoking status, socioeconomic status, and assessment center. Adjustment for other smoking metrics yielded similar results. Lung cancer risk increased per 1 unit decrease in FEV1 (HR=1.80, 95% CI: 1.64-1.98, p=3.3×10-34), FVC (HR=1.45, 1.30-1.60, p=2.3×10-12), and FEV1/FVC (HR=1.39, 1.33-1.46, p=1.3×10-38). This pattern was observed for adenocarcinoma (n=300): FEV1 (HR=1.77, p=6.0×10-12), FVC (HR=1.48, p=1.4×10-5), FEV1/FVC (HR=1.34, p=8.3×10-11); and squamous cell carcinoma (n=166): FEV1 (HR=1.97, p=9.9×10-10), FVC (HR=1.60, p=1.0×10-4), FEV1/FVC (HR=1.38, p=5.9×10-8). Next, a genome-wide association study of 67,708 UKB participants and 12.6 million variants was carried out to develop genetic IVs for LF. Results were filtered to retain independent variants (R2<0.2) associated with each LF phenotype (p<5×10-8). The following IVs were developed: FEV1 (n=28 variants, 0.72% of variation explained), FVC (n=44, 1.08%), and FEV1/FVC (n=45, 1.85%). Odds ratios (OR) for each IV and lung cancer were estimated for 18,686 cases 15,190 controls (>80% European ancestry) from 23 studies. Effect estimates were combined using maximum-likelihood MR models to estimate causal ORs. MR results indicate that genetic scores associated with improved airflow are unrelated to lung cancer risk: FEV1 (OR=1.00, 95% CI: 0.96-1.03, p=0.86), FVC (OR=1.00, 0.97-1.03, p=0.93) and FEV1/FVC (OR=1.00, 0.91-1.10, p=0.95). The null association observed for the genetic determinants of FEV1, FVC and FEV1/FVC was not modified by tumor histology or smoking status. Conclusions: LF is a robust predictor of lung cancer risk, however, our findings do not support the existence of causal pathways that are independent of obstructive lung disease or smoking. This apparent lack of a causal relationship should be interpreted with caution since pleiotropic effects of LF loci cannot be ruled out. Citation Format: Linda Kachuri, Mattias Johansson, Paul Brennan, Phillip Haycock, Geoffrey Liu, Maria Teresa Landi, David C. Christiani, Neil E. Caporaso, Xifeng Wu, Melinda C. Aldrich, Demetrius Albanes, Adonina Tardón, Gad Rennert, Chu Chen, Gary E. Goodman, Jennifer A. Doherty, Heike Bickeböller, Dawn Teare, Lambertus A. Kiemeney, Stig E. Bojesen, John K. Field, Aage Haugen, Stephen Lam, Loic Le Marchand, Matthew B. Schabath, Angeline S. Andrew, Jonas Manjer, Philip Lazarus, Susanne M. Arnold, Valérie Gaborieau, Richard Martin, Caroline Relton, George Davey Smith, Christopher I. Amos, James D. McKay, Rayjean J. Hung. Lung function and lung cancer risk: a Mendelian randomization study of UK Biobank cohort and the International Lung Cancer Consortium [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 2292. doi:10.1158/1538-7445.AM2017-2292

Characterizing Phenotypic Abnormalities Associated with High-Risk Individuals Developing Lung Cancer Using Electronic Health Records from the All of Us Researcher Workbench

Abstract 7331: Establishing a single-cell eQTL dataset of lung tissues from Asian never-smokers to identify cell-type specific lung cancer susceptibility genes

Abstract LB-298: the Use of Lung Cancer Risk Prediction Model to Stratify Heavy Smokers: the Power of Routinely Collected Lab Data

Performance of a Machine Learning Algorithm Using Electronic Health Record Data to Identify and Estimate Survival in a Longitudinal Cohort of Patients With Lung Cancer

Identification of susceptibility pathways for the role of chromosome 15q25.1 in modifying lung cancer risk

Abstract 3483: Epigenome-wide association study of lung cancer among never-smokers in two prospective cohorts in Shanghai

Abstract 2292: Lung Function and Lung Cancer Risk: a Mendelian Randomization Study of UK Biobank Cohort and the International Lung Cancer Consortium

Abstract LB136: Global metabolomics identified promising biomarkers for lung cancer risk among former smokers

Development of Lung Cancer Risk Prediction Machine Learning Models for Equitable Learning Health System: Retrospective Study

Trends in Subpopulations at High Risk for Lung Cancer

Risk Factors for Lung Cancer in an Underrepresented Safety-Net Screening Cohort

High-throughput characterization of functional variants highlights heterogeneity and polygenicity underlying lung cancer susceptibility

Genomic Insights for Personalized Care: Motivating At-Risk Individuals Toward Evidence-Based Health Practices

Identification of risk loci and a polygenic risk score for lung cancer: a large-scale prospective cohort study in Chinese populations

Development and Evaluation of a Polygenic Risk Score for Lung Cancer in Never-Smoking Women: A Large-Scale Prospective Chinese Cohort Study

Curating retrospective multimodal and longitudinal data for community cohorts at risk for lung cancer

Metabolic Obesity Phenotypes and Risk of Lung Cancer: A Prospective Cohort Study of 450,482 UK Biobank Participants

Large-scale Association Analysis Identifies New Lung Cancer Susceptibility Loci and Heterogeneity in Genetic Susceptibility Across Histological Subtypes

Phenome-wide Mendelian randomisation analysis of 378,142 cases reveals risk factors for eight common cancers

Validation of a Deep Learning-Based Model to Predict Lung Cancer Risk Using Chest Radiographs and Electronic Medical Record Data

Stratifying Lung Adenocarcinoma Risk with Multi-ancestry Polygenic Risk Scores in East Asian Never-Smokers