Contributions of smoking behaviour to lung cancer subtypes: A two-sample Mendelian randomization study

Our plain-English summary of Contributions of smoking behaviour to lung cancer subtypes: A two-sample Mendelian randomization study, by M.A. Jareebi, A.Y. Alqassim (Medicine, 2026).

The study

• Authors: M.A. Jareebi, A.Y. Alqassim
• Journal: Medicine
• Published: 2026;105(42):e51170
• Read it: PubMed · Publisher · Free full text

Abstract

Smoking is an established cause of lung cancer (LC), but the relative causal contributions of whether a person ever smokes and how heavily they smoke across histological subtypes remain uncertain. This study investigated the causal effects of smoking status and smoking intensity on major LC subtypes. Two-sample Mendelian randomization (MR) was performed. Genetic instruments were drawn from the GWAS and Sequencing Consortium of Alcohol and Nicotine Use: 85 variants for smoking status, a binary trait contrasting ever with never smokers, and 23 variants for smoking intensity, a continuous trait measuring cigarettes smoked per day. Outcomes were 5 LC phenotypes from FinnGen: non-small cell LC (1627 cases), adenocarcinoma (571), squamous cell carcinoma (365), small cell LC (179), and mesothelioma (133). Inverse-variance weighted regression with multiplicative random effects was the primary estimator, with weighted median, weighted mode, and MR-Egger as sensitivity analyses. Benjamini-Hochberg correction was applied across all ten tests. Analyses used R 4.2.3 with TwoSampleMR 0.7.9. Smoking intensity was causally associated with all 4 smoking-related subtypes: squamous cell carcinoma (odds ratio [OR]: 5.95 per standard deviation increase, 95% confidence interval [CI]: 3.12-11.37), small cell LC (OR: 4.43, 95% CI: 1.60-12.32), non-small cell LC (OR: 3.45, 95% CI: 2.30-5.18), and adenocarcinoma (OR: 2.66, 95% CI: 1.40-5.06). All 4 survived correction for multiple testing and were reproduced by the weighted median and weighted mode estimators. Smoking status showed positive point estimates for all 4 subtypes (OR: 1.40-1.61) but reached only nominal significance for non-small cell LC (OR: 1.51, 95% CI: 1.04-2.19, P = .030), which did not survive correction and was not reproduced by other estimators. Neither exposure was associated with mesothelioma. Smoking intensity shows robust causal effects on all smoking-related LC subtypes, whereas the evidence for smoking status is directionally consistent but considerably weaker. The null finding for mesothelioma, a cancer principally caused by asbestos, supports the specificity of the instruments. These results indicate that reducing consumption among established smokers warrants emphasis alongside preventing uptake.

Copyright © 2026 the Author(s). Published by Wolters Kluwer Health, LLC.

In plain English

This Mendelian randomization study used genetic variants to estimate smoking's effects on lung cancer types. GSCAN data supplied 85 variants for smoking status (ever versus never smoked regularly) and 23 for smoking intensity (cigarettes per day). Outcomes came from FinnGen, a Finnish biobank (European ancestry): non-small cell lung cancer (1627 cases), adenocarcinoma (571), squamous cell carcinoma (365), small cell lung cancer (179) and mesothelioma (133). Smoking intensity was associated with all four lung cancer types. Odds ratios per standard deviation increase in cigarettes per day ranged from 2.66 (95% CI 1.40 to 5.06) for adenocarcinoma to 5.95 (95% CI 3.12 to 11.37) for squamous cell carcinoma, all surviving correction for multiple testing. Smoking status, expressed per unit increase in genetically predicted log-odds of ever smoking, was nominally significant only for non-small cell lung cancer (odds ratio 1.51, 95% CI 1.04 to 2.19), not surviving correction. Neither exposure was associated with mesothelioma, which the authors describe as principally caused by asbestos. The authors concluded smoking intensity had robust causal effects, but noted limited power for rarer subtypes, including mesothelioma.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: Two-sample Mendelian randomization study using published genetic summary statistics, with genetic associations for exposures and outcomes taken from separate datasets
• Who was studied: Genetic summary data. Exposure data came from GSCAN; smoking status was based on 632,802 people. Outcome data came from FinnGen, a Finnish nationwide biobank of people of European ancestry: non-small cell lung cancer 1627 cases and 217,165 controls; adenocarcinoma 571 cases and 218,221 controls; squamous cell carcinoma 365 cases and 218,427 controls; small cell lung cancer 179 cases and 218,613 controls; mesothelioma (all other cancers excluded) 133 cases and 174,006 controls
• Exposure: Genetically predicted smoking, not asbestos. Smoking status (ever smoked regularly versus never; 85 variants) was expressed per unit increase in the log-odds of ever smoking. Smoking intensity (cigarettes smoked per day; 23 variants) was expressed per standard deviation increase in cigarettes per day
• Main results:
  • Smoking intensity and squamous cell carcinoma: OR 5.95 (95% CI 3.12 to 11.37), P less than .001, the strongest effect found
  • Smoking intensity and small cell lung cancer: OR 4.43 (95% CI 1.60 to 12.32)
  • Smoking intensity and non-small cell lung cancer: OR 3.45 (95% CI 2.30 to 5.18). Smoking intensity and adenocarcinoma: OR 2.66 (95% CI 1.40 to 5.06). All four smoking intensity associations remained significant after Benjamini-Hochberg correction
  • Smoking status, expressed per unit increase in the log-odds of ever smoking, gave positive point estimates for the four lung cancer subtypes, from OR 1.40 (squamous cell carcinoma) to OR 1.61 (small cell lung cancer)
  • Smoking status reached only nominal significance for non-small cell lung cancer (OR 1.51, 95% CI 1.04 to 2.19, P = .030). This did not survive correction for multiple testing (adjusted P = .059)
  • Mesothelioma and smoking intensity: OR 1.42 (95% CI 0.57 to 3.53, P = .454), not associated
  • Mesothelioma and smoking status: OR 3.07 (95% CI 0.82 to 11.50, P = .096), not statistically significant. The authors reported wide confidence intervals, reflecting the 133 mesothelioma cases
• The authors' conclusion: The authors concluded that smoking intensity shows robust causal effects on all smoking-related lung cancer subtypes, and that the evidence for smoking status is directionally consistent but considerably weaker. They stated that the null mesothelioma finding, a cancer they describe as principally caused by asbestos, supports the specificity of the genetic instruments
• Limitations the authors state:
  • Both data sources were of European ancestry, which limits generalizability to other populations
  • The analysis was underpowered for rarer subtypes, particularly small cell lung cancer and mesothelioma. Null findings for smoking status should not be read as evidence of no effect
  • The mesothelioma comparison included only 133 cases and had wide confidence intervals
  • MR-Egger was not informative for smoking status, so pleiotropy was less thoroughly excluded for that exposure
  • Smoking status and intensity are genetically correlated, and the analysis estimated total rather than direct effects
  • The exposures reflect lifetime genetic liability, not smoking at a particular age

Topics

The National Library of Medicine indexes this study under subjects our site covers: Lung cancer.

Medical terms in this study, explained in our Lexicon: Lung Cancer, Histological Subtype, Adenocarcinoma, Squamous Cell Carcinoma, Mesothelioma, Odds Ratio, confidence interval.

Related studies

• Integrative eQTL and Mendelian randomization analysis reveals key genetic markers in mesothelioma — J. Li, X. Wang, et al. (2025)
• Obesity and risk for respiratory diseases: a Mendelian randomization study — W. Yang, Y. Yang, et al. (2023)
• Bidirectional Mendelian randomization and potential mechanistic insights into the causal relationship between gut microbiota and malignant mesothelioma — Y. Zhou, H. Zhu, et al. (2025)
• No causal association between pneumoconiosis and three inflammatory immune diseases: a Mendelian randomization study — Y.J. Du, Z.W. Lu, et al. (2024)
• Identification of Novel Biomarkers and Potential Therapeutic Targets for Systemic Sclerosis: An Integrated Analysis of Plasma Proteome-Wide Mendelian Randomization and Transcriptome — H. Li, Q. Li, et al. (2025)
• Cigarette smoking, asbestos exposure, lung cancer, and sample size — M.F. Petrini (1998)

Browse more research in Medicine, or return to medical research.