Screening performance in high-risk groups of breast cancer by integrating classical risk factors, mammographic density and polygenic risk

Abstract Background: Risk prediction models integrating classical risk factors (CRF), mammographic density (MD), and polygenic risk score (PRS) are increasingly developed to identify high-risk groups of breast cancer. Few studies investigate the screening performance in high-risk groups by these models.Methods: Based on a median follow-up of 11.7 years of 7794 women from the Multi-modality Independent Screening Trial (MIST), four risk prediction models with CRF (ModelCRF), CRF and MD (ModelCRF+MD), CRF and PRS (ModelCRF+PRS), and all three components (ModelFULL) were developed to identify high-risk groups of breast cancer. The hazard ratio (HR) and 95% confidential interval (CI) of breast-cancer mortality for high-risk groups compared to low-risk groups was calculated to determine potential benefit of risk-reducing interventions. The detection rate (DR), accuracy and cancer-stage for clinical breast examination (CBE), breast ultrasonography (BUS), and mammography (MAM) were compared to determine the optimal screening method for high-risk groups.Results: The areas under the curve of risk prediction model increased from 0.573 (95%CI: 0.532-0.614) for ModelCRF, to 0.587 (95%CI: 0.544-0.630) for ModelCRF+MD, 0.670 (95%CI: 0.622-0.717) for ModelCRF+PRS and 0.674 (95%CI: 0.623-0.725) for ModelFULL. The HRs of breast cancer mortality for high-risk groups compared to low-risk groups increased from 1.81 (95%CI: 1.17-2.81) for ModelCRF, to 2.22 (95%CI: 1.35-3.67) for ModelCRF+MD, 2.48 (95%CI: 1.43-4.29) for ModelCRF+PRS, and 3.68(95%CI: 1.94-6.99) for ModelFULL. Among high-risk groups by ModelCRF, the DR of BUS was similar to that of MAM (3.926/1,000 vs. 2.399/1,000, P=0.193), but significantly higher than that of CBE (1.091/1,000, P=0.024). Compared with MAM, BUS showed significantly lower sensitivity (50.0% vs. 81.8%, P=0.026), but comparable specificity (99.2% vs. 99.3%), positive prediction values (22.9% vs. 34.6%), and negative prediction values (99.8% vs. 99.9%). Further analyses showed no significant difference in the proportions of early-stage breast cancer detected between BUS and MAM (50.00% vs. 61.54%, P=0.673). Similar results were observed in high-risk groups by other models.Conclusions: Accurate risk assessment integrating CRF, MD and PRS is needed to identify high-risk groups of breast cancer. The higher the risk, the greater the benefit of the intervention. BUS was comparable to MAM for screening breast cancer in high-risk groups.