The current state of polygenic scores for the development of lung cancer: a systematic review and validation in UK Biobank

TL;DR

This study reviews and validates lung cancer polygenic scores in the UK Biobank, finding that most scores are weak predictors and perform better in tobacco users.

Contribution

The study systematically evaluates and compares the performance of 60 lung cancer polygenic scores in a large population cohort.

Findings

Most evaluated polygenic scores showed a hazard ratio per standard deviation greater than 1.1.

22 out of 39 scores had a C-index greater than 0.55, indicating moderate discrimination.

Performance varied by tobacco use, with most scores performing better in tobacco users.

Abstract

Risk-stratified lung cancer screening programs identify high-risk individuals who use tobacco but do not account for underlying genetic susceptibility. Many polygenic scores (PGS) have been developed for lung cancer, but it is unclear which, if any, are suitable for identifying high-risk individuals in the general population.

We used a systematic review to identify published lung cancer PGS, which were implemented and validated in the UK Biobank (UKB) cohort. Performance (discrimination and accuracy) was compared. Subgroup analyses by sex, ethnicity, and smoking status identified differences across the population.

We identified 60 lung cancer PGS published since 2012. Most scores were associated with lung cancer risk in UKB. Of the 39 evaluated PGS, 33 had a hazard ratio per standard deviation greater than 1.1 and 22 had a C-index greater than 0.55. Most PGS perform better in individuals who use tobacco than those who do not, although for a small number of scores (n = 8) the reverse is true.

Performance of lung cancer PGS is weak compared to scores for other cancers; the potential benefit of combining genetics with other risk factors for lung cancer remains unclear. Selection of a suitable score is context dependent and requires consideration of the characteristics of the target population (such as ethnicity and tobacco usage).

Figures