Helfand Talia Y, Wei Jun, Mulford Ashley J, Tran Huy, Shi Zhuqing, Wang Chi-Hsiung, Rifkin Andrew S, Ashworth Annabelle, Zheng S Lilly, Helfand Brian T, Dunnenberger Henry M, Duggan David, Hulick Peter J, DePersia Allison, Sanders Alan R, Xu Jianfeng
Program for Genomic Translational Research, NorthShore University HealthSystem (Endeavor Health), Evanston, IL.
Genomic Health Initiative, Endeavor Health, Evanston, IL.
JCO Precis Oncol. 2025 Jul;9:e2500081. doi: 10.1200/PO-25-00081. Epub 2025 Jul 2.
Family history (FH) and pathogenic variants (PVs) in Lynch syndrome (LS) genes are established risk factors of colorectal cancer (CRC). This study evaluates whether newly published polygenic scores (PGSs) improve CRC prediction of known risk factors.
The associations of CRC risk with FH, PVs in LS genes (, , , ), and PGS were investigated in the UK Biobank (UKB) using population attributable risk percentage (PAR%) and Cox regression. Model performance was assessed using Concordance Index (C-index) in UKB and validated in the Genomic Health Initiative (GHI), a US health care biobank with more diverse ancestry populations.
In UKB, 18.99%, 11.43%, and 0.42% of participants were positive for PGS, FH, and PVs, respectively, with corresponding PAR% of 29.97%, 6.27%, and 1.25%. In multivariable analysis adjusting for age, sex, and genetic background, each genetic factor independently predicted CRC risk ( < .001). Significant interactions between PVs and FH ( < .001) and PVs and PGS ( = .04) were found; hazard ratio (95% CI) of PVs was significantly higher in those with FH (7.90 [6.27 to 9.94]) than without FH (2.89 [2.32 to 3.60]) and that of PGS was significantly lower in PV carriers (1.69 [1.18 to 2.42]) than noncarriers (2.85 [2.73 to 2.98]). The combined genetic model had a significantly higher C-index (0.677) than models with PVs (0.573), FH (0.578), and PGS (0.669) alone ( < .001). This enhanced performance was validated in GHI across European and non-European ancestries.
PGS complements FH and PVs of LS genes in predicting CRC risk and improves prediction performance beyond traditional genetic factors across diverse populations.
林奇综合征（LS）基因中的家族史（FH）和致病变异（PVs）是结直肠癌（CRC）已确定的风险因素。本研究评估新发表的多基因评分（PGSs）是否能改善对已知风险因素的CRC预测。
在英国生物银行（UKB）中，使用人群归因风险百分比（PAR%）和Cox回归研究CRC风险与FH、LS基因中的PVs（、、、）以及PGS之间的关联。在UKB中使用一致性指数（C-index）评估模型性能，并在基因组健康倡议（GHI）中进行验证，GHI是一个拥有更多样化祖先群体的美国医疗生物银行。
在UKB中，分别有18.99%、11.43%和0.42%的参与者PGS、FH和PVs呈阳性，相应的PAR%分别为29.97%、6.27%和1.25%。在对年龄、性别和遗传背景进行调整后的多变量分析中，每个遗传因素均独立预测CRC风险（<0.001）。发现PVs与FH（<0.001）以及PVs与PGS（=0.04）之间存在显著相互作用；有FH的人群中PVs的风险比（95%CI）（7.90[6.27至9.94]）显著高于无FH的人群（2.89[2.32至3.60]），而PV携带者中PGS的风险比（1.69[1.18至2.42]）显著低于非携带者（2.85[2.73至2.98]）。联合遗传模型的C-index（0.677）显著高于单独使用PVs（0.573）、FH（0.578）和PGS（0.669）的模型（<0.001）。这种增强的性能在GHI中针对欧洲和非欧洲血统人群得到了验证。
PGS在预测CRC风险方面补充了LS基因的FH和PVs，并在不同人群中提高了超越传统遗传因素的预测性能。
