Challenges in optimization of breast cancer screening
Challenges in optimization of breast cancer screening
Current challenges in breast cancer screening optimization include ductal carcinoma in situ (DCIS), overdiagnosis, and screening context. For DCIS, large gaps in knowledge on natural history exist and screening programs are optimized towards invasive breast cancer (IBC). Furthermore, overdiagnosis estimates range from 0-54% for total breast cancer and 20-91% for DCIS. Also, screening context differs, due to population characteristics, available resources, and policy.
The main objectives in this thesis of Keris Poelhekken were to (1) improve understanding of DCIS, (2) explain the variation in overdiagnosis, and (3) evaluate screening effectiveness in various contexts.
We found that simulation models differ widely in assumptions on DCIS natural history, with 8-90% progressive DCIS and 0-10% regression. We developed a DCIS model with estimates of 20% DCIS progressing to IBC and 5% regressing in Dutch context. Furthermore, the range in overdiagnosis estimates can be explained by definition (estimates of 20-94%), short follow-up time (7% overestimation), and screening strategy (16% overestimation). Proper overdiagnosis estimates need a clear definition (including screen-, clinically detected, and progression to IBC), long follow-up (10 years for IBC and 20 years for DCIS), to include DCIS and IBC separately, appropriate methods, and transparent reporting. Breast cancer models should adhere to modelling guidelines to ensure reliable estimates. In good adherence with modelling guidelines, we found 20% overdiagnosis of DCIS in Dutch context and 2.4% of IBC in Flemish context. Models, including SimRisc and SimDCIS, are useful to inform policy makers and women of screening age. Improved models and overdiagnosis estimates will allow screening optimization for specific contexts.