Using clinical and genetic risk factors for risk prediction of 8 cancers in the UK Biobank

Sep 21, 2026

Uploaded by Anda

Background: Models that combine polygenic risk scores with clinical factors to predict cancer-specific risk have been constrained by polygenic risk score derivation choices and by incomplete selection of clinical variables. Methods: Using UK Biobank, the study trained optimal polygenic risk scores for eight cancers and selected clinical variables from 733 baseline traits via extreme gradient boosting (XGBoost). Cox proportional hazards models combined genetic and clinical information to predict risk. Results: High accuracy was achieved across eight cancers, enabling identification of individuals at elevated risk, including strong relative risk differences between top and low score quantiles and different patterns for early- versus late-onset groups. Conclusion: Combined genetic-Clinical modeling showed overall stronger predictive performance than genetic-only or clinical-only approaches, with greater predictive value for early-onset patients under the polygenic risk score model and for late-onset patients under the clinical risk model.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Abstract

Background

Methods

Results

Conclusion

Cancer risk context

Genetic and non-genetic factors

Polygenic risk scores and clinical interactions

Study limitations and motivation

JNCI Cancer Spectrum, 2024, 8(2), pkae008 [https://doi.org/10.1093/jncics/pkae008](https://doi.org/10.1093/jncics/pkae008) Advance Access Publication Date: February 14, 2024 Article Using clinical and genetic risk factors for risk prediction of 8 cancers in the UK Biobank Jiaqi Hu , BA,1 Yixuan Ye, PhD,2 Geyu Zhou, PhD,2 Hongyu Zhao , PhD2,3,􀀃 1Department of Chronic Disease Epidemiology, Yale School of Public Health, New Haven, CT, USA 2Program of Computational Biology and Bioinformatics, Yale University, New Haven, CT, USA 3Department of Biostatistics, Yale School of Public Health, New Haven, CT, USA 􀀃 Correspondence to: Hongyu Zhao, PhD, Department of Biostatistics, Yale School of Public Health, PO Box 208034, 60 College St, New Haven, CT, USA 06520 ([e-mail: hongyu.zhao@yale.edu](e-mail: hongyu.zhao@yale.edu)) . Abstract Background: Models with polygenic risk scores and clinical factors to predict risk of different cancers have been developed, but these models have been limited by the polygenic risk score–derivation methods and the incomplete selection of clinical variables. Methods: We used UK Biobank to train the best polygenic risk scores for 8 cancers (bladder, breast, colorectal, kidney, lung, ovarian, pancreatic, and prostate cancers) and select relevant clinical variables from 733 baseline traits through extreme gradient boosting (XGBoost) . Combining polygenic risk scores and clinical variables, we developed Cox proportional hazards models for risk prediction in these cancers. Results: Our models achieved high prediction accuracy for 8 cancers, with areas under the curve ranging from 0.618 (95% confidence interval ¼ 0.581 to 0.655) for ovarian cancer to 0.831 (95% confidence interval ¼ 0.817 to 0.845) for lung cancer. Additionally, our models could identify individuals at a high risk for developing cancer. For example, the risk of breast cancer for individuals in the top 5% score quantile was nearly 13 times greater than for individuals in the lowest 10% . Furthermore, we observed a higher proportion of individuals with high polygenic risk scores in the early-onset group but a higher proportion of individuals at high clinical risk in the late-onset group. Conclusion: Our models demonstrated the potential to predict cancer risk and identify high-risk individuals with great generalizability to different cancers. Our findings suggested that the polygenic risk score model is more predictive for the cancer risk of early-onset patients than for late-onset patients, while the clinical risk model is more predictive for late-onset patients. Meanwhile, combining polygenic risk scores and clinical risk factors has overall better predictive performance than using polygenic risk scores or clinical risk factors alone. Cancer is a widespread disease that accounts for millions of deaths each year. It is estimated that nearly 10 million deaths were caused by cancer in 2020 (1) . The etiology of cancer can be attributed to genetic and nongenetic factors. In recent years, large-scale genome-wide association studies (GWASs) have identified thousands of single-nucleotide variations (SNVs—formerly SNPs) associated with cancer (2-7) . For example, a recent GWAS combining 100 204 colorectal cancer (CRC) cases and 154 587 controls of European and East Asian ancestry through meta-analysis successfully identified 205 independent SNVs strongly associated with CRC (4), demonstrating the role of genetic factors in cancer development. Nongenetic factors, such as increased age (8) and smoking (9) are widely known as general risk factors of cancer. Additionally, studies have identified risk factors specific to certain cancers. For instance, reproductive factors, such as the use of exogenous hormones, have been associated with an increased risk of breast cancer (10) . Polygenic risk scores have been proposed to accurately predict the risk of developing cancer by aggregating the genetic effects of SNVs (11-13) . The comprehens

FAQs

Which data source and modeling approach were used to build the risk prediction models?

The models were trained using UK Biobank. Extreme gradient boosting (XGBoost) was used to select relevant clinical variables from baseline traits, and Cox proportional hazards models were used to combine polygenic risk scores with clinical variables.

How well did the models predict risk across the eight cancers?

Prediction accuracy was high for all eight cancers, with areas under the curve ranging from 0.618 for ovarian cancer to 0.831 for lung cancer.

What did the study find about identifying high-risk individuals and differences between early- and late-onset?

The models could identify individuals at high risk, including large relative risk differences between high and low score quantiles. The study observed a higher proportion of individuals with high polygenic risk scores in early-onset groups, while individuals with high clinical risk were more frequent in late-onset groups.

Sep 21, 2026

Uploaded by Anda