Article Text
PDF
Abstract
We investigated phenotypic leucocyte telomere length (LTL), genetically predicted LTL (gTL), and lung cancer risk among 371 890 participants, including 2829 incident cases, from the UK Biobank. Using multivariable Cox regression, we found dose-response relationships between longer phenotypic LTL (p-trendcontinuous=2.6×10-5), longer gTL predicted using a polygenic score with 130 genetic instruments (p-trendcontinuous=4.2×10-10), and overall lung cancer risk, particularly for adenocarcinoma. The associations were prominent among never smokers. Mendelian Randomization analyses supported causal associations between longer telomere length and lung cancer (HRper 1 SD gTL=1.87, 95% CI: 1.49 to 2.36, p=4.0×10-7), particularly adenocarcinoma (HRper 1 SD gTL=2.45, 95%CI: 1.69 to 3.57, p=6.5×10-6).
- Lung Cancer
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Introduction
Telomere length reflects the cumulative burden of exposures, endogenous factors, and age. We investigated phenotypic leucocyte telomere length (LTL), telomere length predicted using a polygenic score (PGS; gTL), and lung cancer risk in the United Kingdom (UK) Biobank. Further, we examined the associations by histology and subgroups defined by smoking status and sex. To evaluate causal relationships, we conducted Mendelian Randomization (MR) analyses using 130 genetic instruments that predict longer LTL.1 2 Our study could help identify high-risk subpopulations that do not have but may benefit from lung cancer screening.
Materials and methods
Study population
The UK Biobank is described previously3 and in the Supplementary Materials. Among 502 409 participants at baseline, we excluded 372 subjects with discrepancies between self-reported and genetic sex; 46 577 with any cancer diagnosis; 57 253 with respiratory diseases; and 4779 with hematologic/immunologic disorders. Our analytic dataset had 393 428 participants.
Follow-up time started at the visit date to the assessment centres in 2006–2010 and ended at the date of primary incident lung cancer diagnosis, death, or administrative censoring. Lung cancer diagnosis was defined by International Classification of Diseases 10th revision (ICD-10) codes C34.0-C34.9. ICD-O-3 code 8140 defined adenocarcinoma (LUAD), while 8052, 8084, 8073, and 8083 defined squamous cell carcinoma (SCC).
LTL measurements
The DNA extraction, multiplex quantitative polymerase chain reaction (qPCR) assay, and quality control procedures were previously described.4 The ratio of telomere (T) to single-copy gene (S) copy number (‘T/S ratio’) reflects the average telomere abundance across all chromosomes in leucocytes of an individual, which was further adjusted for batch variation (‘adjusted relative T/S ratio’).4
PGS for longer gTL and MR analyses
Using a one-sample approach,1 we constructed a weighted PGS that predicted longer gTL with 130 single nucleotide polymorphisms (SNPs) identified among Europeans in the UK Biobank. These SNPs explained 4.54% of the variance in LTL.1 We note that LTL is affected by the exposome, which can influence LTL variation but unlikely bias the identification of LTL-related SNPs. The SNPs were linkage disequilibrium pruned (r2 > 0.01), reached genome-wide significance (p<8.31×10−9), and included in previous MR analyses.1 The weights were based on reported Z-standardised beta-estimates.1 To estimate causal associations between gTL and lung cancer and its subtypes among Europeans, we conducted MR analyses using the same 130 SNPs with MR-PRESSO in R. The HR from MR analyses reflect increased lung cancer risk per one SD increase in gTL.1
Statistical analyses
Cox regression was used to estimate HRs and 95% CI of incident lung cancer in relation to quartiles (Q) of phenotypic LTL (‘adjusted relative T/S Ratio’; Q1:<0.74; Q2:0.74 to<0.82; Q3:0.82 to<0.91; Q4:≥0.91), adjusted for age, sex, race/ethnicity, detailed smoking history/intensity, assessment centre, body mass index, Townsend Deprivation Index, alcohol intake, secondhand smoke exposure, and leucocyte differentials. Follow-up time was the timescale. LUAD and SCC were analysed separately. We fitted separate Cox models with gTL as the main effect among Europeans, without leucocyte differential adjustment.
Linear trends were estimated using continuous gTL and log-transformed phenotypic LTL. Stratified analyses were conducted among subgroups defined by smoking status (never, former, current) and sex. We tested for multiplicative effect modification of LTL-lung cancer associations by smoking status and sex using cross-product terms. Heterogeneity between HR estimates was assessed using Z-score sign tests.
Results
Phenotypic LTL and lung cancer
There were 2829 incident lung cancer cases, including 1078 LUAD and 487 SCC, diagnosed over the 12.36±1.64 SD year follow-up. The average time between LTL measurements and lung cancer diagnosis was 6.54±3.22 SD years. We found a strong dose-response relationship between longer phenotypic LTL and increased lung cancer risk (p-trendcontinuous=2.6×10-5), particularly LUAD (p-trendcontinuous=6.6×10-10) but not SCC (table 1; p-differenceLUAD vs. SCC=6.7×10-5). Findings were similar among Europeans and when including participants with chronic respiratory diseases at enrollment (online supplemental table 1).
Supplemental material
Associations between measured phenotypic leucocyte telomere length and lung cancer risk in 371 890 participants from the UK Biobank
We observed multiplicative effect modification of the LTL-lung cancer association by smoking, with attenuation among former (p-interaction=0.008) and current smokers (p-interaction=0.008) vs never smokers. We did not detect an interaction between phenotypic LTL and sex (p-interaction=0.909). Among subgroups, we found associations between longer phenotypic LTL and increased lung cancer risk among never smokers (p-trend=3.3×10-7), including never-smoking women (p-trend=4.0×10-5) and men (p-trend=2.0×10-3), as well as former-smoking women (p-trend=5.7×10-4) (table 2). When analysing LUAD, the trends were consistent with overall lung cancer (online supplemental table 2). Due to sparse data, we could not analyse SCC among never smokers and consistent associations were not detected in other subgroups (online supplemental table 2)
Associations between measured phenotypic leucocyte telomere length and overall lung cancer risk in the UK Biobank among subgroups defined by sex and smoking status
PGS and MR analyses among Europeans
In PGS analyses, we found strong dose-response relationships between longer gTL and lung cancer risk (p-trend=4.2×10-10), particularly LUAD (p-trend=2.1×10-8) but not SCC (figure 1). We did not detect multiplicative interactions between gTL and smoking or sex (p-interactions>0.05).
Genetically predicted leucocyte telomere length and risk of lung cancer and its histological subtypes. Using European data from the UK Biobank, we constructed a weighted PGS that predicts longer LTL. This PGS was based on 130 out of 197 genetic variants identified in the UK Biobank that reached genome-wide significance (p<8.31×10−9). These genetic variants were included in the previous MR analyses and were LD pruned (R2 > 0.01) (Codd et al. Nat Genet 2021). The weights were based on the previously reported Z-standardised beta estimates for SNP-LTL associations (Codd et al. Nat Genet 2021). We categorised the PGS into quartiles and analysed associations with risk of lung cancer, LUAD, and SCC among Europeans using multivariable Cox regression adjusted for sex, body mass index, age at recruitment, detailed smoking history/intensity, assessment centre, Townsend Deprivation Index, alcohol use, and secondhand smoke exposure. Linear trends were estimated by analysing continuous PGS.
In MR analyses, we found evidence for causal associations between longer gTL and lung cancer (HRper 1 SD gTL=1.87, 95% CI: 1.49 to 2.36, p=4.0×10-7), particularly LUAD (HRper 1 SD gTL=2.45, 95% CI: 1.69 to 3.57, p=6.5×10-6) but not SCC (HRper 1 SD gTL=1.19, 95% CI: 0.74 to 1.93, p=0.47; p-differenceLUAD vs. SCC=0.02). No horizontal pleiotropy was detected for any SNP.
Discussion
We confirm and expand on previous studies with important findings among subpopulations defined by smoking and sex. Previously, we reported associations between longer phenotypic LTL and increased lung cancer risk when pooling three nested case-control studies, particularly for LUAD among Europeans.5 Those findings were corroborated in separate studies of Chinese6 and heavy-smoking European populations.7 Our PGS and MR analyses were in agreement with studies that found associations between longer gTL and increased lung cancer risk.8–10 In particular, our MR analyses support causal associations between gTL and lung cancer. The magnitude of our gTL and phenotypic LTL effects were similar; suggesting that the genetic component of LTL captured by gTL is an effective surrogate marker, despite explaining a small proportion of variance. Our findings support longer pre-diagnostic LTL in processes relevant to lung carcinogenesis, prominently among non-smokers. Longer LTL potentially reflects decreased senescence or higher replicative potential in pre-cancerous cells, which allows accumulation of genetic alterations that initiate or promote carcinogenesis.
Supplemental material
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by North West - Haydock Research Ethics Committee 3rd Floor - Barlow House 4 Minshull Street Manchester M1 3DZ Telephone: 02071048103. Title of the Database: UK Biobank: a large scale prospective epidemiological resource. Designated Individual: Mrs. Samantha Welsh. REC reference: 21/NW/0157IRAS. project ID: 299116 Participants gave informed consent to participate in the study before taking part.
Acknowledgments
We thank Sharon Savage (NCI) for valuable expertise on telomere biology. Additionally, we thank Richard Cawthon for developing the original qPCR assay to measure relative telomere length and for nearly two decades of collaboration. Additionally, we would like to thank Lisa Finkelstein (NCI) for leading efforts to establish the international material transfer agreements.
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Twitter @MohammadLRahma2
Contributors JYYW wrote the manuscript, analyzed data, and led the study. BB, AKH, JS, WH, and MLR edited the manuscript and analyzed data. MJM and SMG edited the manuscript. NR and QL edited the manuscript and supervised the study.
Funding This study was supported by intramural funding from the National Cancer Institute (NCI), Division of Cancer Epidemiology and Genetics (DCEG).
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.