Clinical Predictors of BRCA1/2 Variants in High-Risk Breast Cancer Patients in China

102 BRCA1/2 pathogenic variants were identified in a cohort of 1,204 high-risk breast cancer patients from Zhejiang Cancer Hospital between 2017 and 2021. Wei Gu developed a multivariable logistic regression model using clinical predictors like histology, molecular subtype, age, and family history. The model achieved an AUC of 0.758 and was validated internally with 500 bootstrap resamples.

Use Cases

• Training a binary classification model to predict BRCA1/2 carrier status based on clinical features like age and family history.
• Benchmarking new predictive models against the reported logistic regression model with an AUC of 0.758.
• Analyzing the association between clinical factors like triple-negative breast cancer (TNBC) and BRCA1/2 variant prevalence.
• Conducting a decision curve analysis to evaluate the clinical utility of genetic testing strategies.

Strengths

• Includes data from 1,204 high-risk breast cancer patients, described as a large cohort for mainland China.
• Model performance is quantitatively reported with an AUC of 0.758, Brier score of 0.071, and bootstrap validation (500 resamples).
• Clinical predictors are clearly defined, including histology, molecular subtype, age, and family history.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the underlying clinical data is unknown, which may limit suitability assessment.
• Data is from a single hospital in eastern China, which may reflect geographic bias.

Provenance

Source

Zhejiang Cancer Hospital, eastern China.

Collection Method

Clinical data collection and targeted NGS for BRCA1/2 on blood samples, with variants classified by ClinVar and ACMG-AMP guidelines.

Time Range

2017–2021

Freshness

Last updated 2026-06-02 04:21:26; freshness should be verified.

Geography

Eastern China