Improving Breast Cancer Risk Prediction in Women of Diverse Genetic Ancestry

Abstract

Background: Polygenic risk scores (PRS) improve personalised breast cancer risk prediction but are derived using predominantly European ancestry data and therefore require ethnicity specific recalibration. People of diverse genetic ancestry have differing Effect Allele Frequencies (EAFs) and research has shown that current European ancestry-derived PRS over-predict their breast cancer risk. Current breast cancer PRS have not been well-validated in young women without a strong family history. Methods: 1) A prospective case-control study on 1000 women (250 cases;750 controls) has been carried out in 30-39-year-olds, who have been risk stratified using a SNP313 PRS. Pathogenic Variants (PVs) in high-risk genes have also been looked for using novel techniques. 2) SNP78 and SNP142 have been calculated and readjusted using Ashkenazi Jewish data in 221 Ashkenazi women (121 cases;100 controls) and validated in 2045 Ashkenazi women (1331 cases;714 controls). 3) SNP142 has been calculated and recalibrations investigated in 138 women (33 cases;105 controls) and validated in 3686 women (1657 cases;2029 controls) of African ancestry. African samples were separated by more specific genetic ancestry and investigated in sub-populations. Results: 1) Mean SNP313 z-score stratifies cases at higher risk than controls (cases 0.30; controls 0.07) although there is overestimation in the ancestrally diverse group. The Global Screening Array (GSA) can identify PVs and has high sensitivity for small base changes (94%) and large rearrangements (87.5%). 2) SNP142 showed good discrimination but overprediction in the Ashkenazi population with mean PRS in cases of 1.54 (95% CI 1.38-1.70) and in controls of 1.20 (95% CI 1.08-1.32). Adjustment using published Ashkenazi data corrected mean risk (cases 1.30 [95% CI 1.16-1.44] and controls 1.02 [95% CI 0.92-1.12]). 3) SNP142 showed poor discrimination and high overprediction in the African population with mean PRS in cases of 1.65 (95% CI 1.23- 2.07) and in controls of 1.61 (95% CI 1.45-1.77). Taking only the African ancestry samples which cluster closely with known African data, readjustment using published African data works well (cases 1.20 [95% CI 0.94-1.46]; controls 1.01 [95% CI 0.94-1.08]). Conclusions: SNP313 PRS can identify young women (30-40-years) at higher breast cancer risk. The GSA can identify breast cancer PVs and could potentially be used in lower-income countries. Women of diverse genetic ancestry have their breast cancer risk over-predicted using unadjusted current PRS. PRSs must be recalibrated using data relevant to the ethnicity in which the PRS is being used. African people should have their ethnicity mapped on a Multi- Dimensional Scaling plot to infer more specific ancestry before any adjustment is made.

Date of Award	24 Jan 2025
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Sacha Howell (Main Supervisor), Miriam Smith (Co Supervisor), Jamie Sergeant (Co Supervisor), Susan Astley Theodossiadis (Co Supervisor) & Gareth Evans (Co Supervisor)