Development and validation of a lung cancer polygenic risk score incorporating susceptibility variants for risk factors

Abstract Incorporating susceptibility genetic variants of risk factors has been reported to enhance the risk prediction of polygenic risk score (PRS). However, it remains unclear whether this approach is effective for lung cancer. Hence, we aimed to construct a meta polygenic risk score (metaPRS) of lung cancer and assess its prediction of lung cancer risk and implication for risk stratification. Here, a total of 2180 genetic variants were used to develop nine PRSs for lung cancer, three PRSs for different histopathologic subtypes, and 17 PRSs for lung cancer‐related risk factors, respectively. These PRSs were then integrated into a metaPRS for lung cancer using the elastic‐net Cox regression model in the UK Biobank ( N = 442,508). Furthermore, the predictive effects of the metaPRS were assessed in the prostate, lung, colorectal, and ovarian (PLCO) cancer screening trial ( N = 108,665). The metaPRS was associated with lung cancer risk with a hazard ratio of 1.33 (95% confidence interval: 1.27–1.39) per standard deviation increased. The metaPRS showed the highest C‐index (0.580) compared with the previous nine PRSs (C‐index: 0.513–0.564) in PLCO. Besides, smokers in the intermediate risk group predicted by the clinical risk model (1.34%–1.51%) with the intermediate‐high genetic risk had a 6‐year average absolute lung cancer risk that exceeded the clinical risk model threshold (≥1.51%). The addition of metaPRS to the clinical risk model showed continuous net reclassification improvement (continuous NRI = 6.50%) in PLCO. These findings suggest the metaPRS can improve the predictive efficiency of lung cancer compared with the previous PRSs and refine risk stratification for lung cancer.