Clinical Scorecard: Design and Justification of the STREAM Trial: A Population-Based Study on Breast Cancer Screening Utilizing Advanced Reader Techniques

At a Glance

Key Highlights

• STREAM is a prospective, non-randomised, parallel-group trial embedded in the Dutch breast cancer screening programme comparing two rounds of DBT screening to DM screening.
• DBT screening aims to improve cancer detection rates and optimize reading time using advanced techniques including AI-aided navigation and modified image acquisition.
• Primary outcomes include participation rate, recall rate, detection rate, positive predictive value, interval cancer rates, advanced cancer detection, and cost-effectiveness.

Guideline-Based Recommendations

Diagnosis

• Use biennial digital mammography (DM) with craniocaudal (CC) and mediolateral-oblique (MLO) views double-read independently by two radiologists.
• In the STREAM trial, DBT with synthetic mammography (SM) and double reading is used to potentially improve detection rates.

Management

• Women with suspicious findings on screening are referred for diagnostic work-up at hospitals.
• DBT screening interpretation strategies include using MLO views without CC, thicker image slabs, AI-aided navigation, and possibly replacing one reader with AI to reduce workload.

Monitoring & Follow-up

• Monitor participation rates, recall rates, detection rates, positive predictive values, interval cancer rates, and advanced cancer detection in screening rounds.
• Evaluate long-term outcomes including breast cancer mortality, overdiagnosis, and cost-effectiveness.

Risks

• Potential increase in recall rates with DBT screening in settings with already low recall rates like the Netherlands.
• Uncertainty regarding impact of DBT on interval cancer rates and cost-effectiveness in the Dutch screening context.

Patient & Prescribing Data

Women aged 49 to 72 invited for breast cancer screening in selected Dutch screening units.

DBT screening involves two-view imaging (CC and MLO) with double reading; participation requires informed consent; exclusion only if wheelchair-bound.

Clinical Best Practices

• Ensure radiologists and radiographers receive specific training on DBT image acquisition and interpretation prior to trial initiation.
• Use double reading with arbitration or consensus to resolve discrepancies in image interpretation.
• Incorporate previous DM images for comparison when available to improve diagnostic accuracy.
• Optimize DBT reading time by employing advanced reading methods such as AI assistance and modified image acquisition protocols.

References

• Dutch breast cancer screening programme details
• European breast cancer screening detection and recall rates
• Digital breast tomosynthesis studies and outcomes
• STREAM trial registration and protocol