Multilevel pathways to esophageal cancer: Population attributable risks of modifiable exposures, surrogate blood biomarkers, and genetic susceptibility, and the mediation effects based on a 10‐year prospective cohort study

Numerous studies have elucidated the contributions of modifiable risk factors, blood-based biomarkers, and genetic variants to the etiology of esophageal cancer (EC). However, it lacks systematic evaluation of their contribution to EC risk in prospective cohorts of the Chinese population. We prospectively examined the association between 10 modifiable risk factors, 54 blood biomarkers, an established 11-locus polygenic risk score (PRS), and EC risk using the Cox model in the Taizhou Longitudinal Study. The participants were enrolled between 2011 and 2014 and followed up to December 31, 2023. Mediation analysis was also conducted to explore the mediation effect of blood biomarkers. Finally, we calculated population attributable fractions (PAFs) to summarize the relative contribution of the three types of risk factors. Participants enrolled between 2018 and 2021 were used for validation. Four modifiable risk factors and 14 tumor/non-tumor blood biomarkers were identified as associated with EC risk. The highest PRS quintile had a 1.59-fold increased risk of EC compared to the combined low-risk group of the bottom 40% population (p = .049). High-density lipoprotein mediated 7.56% of the total effect of alcohol drinking on EC risk, a value of 12.78% in the validation dataset with a borderline significance of p = .054. The combined PAF of EC incidence attributable to high modifiable risk and blood biomarkers was 58.55%, and the PAF increased to 63.35% when PRS was added. Our study offers a comprehensive evaluation of the contribution of modifiable risk factors, clinical blood biomarkers, and genetic factors to EC risk in the Chinese population.