Monday, November 30 | 9:30 a.m. to 10:30 a.m. | M3-SSBR04-1 | Room S402
Researchers will report results from the ScreenTrustMRI trial, which tested whether an AI risk predictor can select the women who would benefit most from supplemental MRI.
Screening mammography lowers mortality but misses some cancers, and most of those can be found with MRI. Supplemental MRI is feasible for only a fraction of screening participants, however.
Fredrik Strand, PhD, MD, from Karolinska University Hospital in Sweden will present his team’s study, which took place within population-based screening at a single hospital.
Negative screening mammograms were analyzed with AISmartDensity, a tool developed through a collaboration between two Swedish universities. Women recalled for mammographic findings were not eligible. Women with high AI-predicted risk were invited to be randomized to supplemental MRI or no MRI.
Of 59,810 women screened, 4,555 (7.6%) were high risk. Of these, 4,103 were eligible and invited, and 1,315 accepted, with 663 assigned to MRI and 652 to no MRI. In this session, attendees will hear that early detection failure occurred in 4.8 per 1,000 women in the MRI group versus 26.1 per 1,000 in the no-MRI group. In the per-protocol population, rates were 3.8 versus 29.3 per 1,000, an 87% risk reduction. Not-high-risk women who were not invited had a rate of 2.8 per 1,000, which did not differ significantly from high-risk women who underwent MRI (p = 0.41).
The authors highlight that the tool makes programmatic supplemental MRI practical, since only about 7% of women would need MRI to achieve a more equitable risk of early detection failure across the screened population.
Find out more by attending this session.
