Abbreviated MRI and contrast-enhanced mammography detect ~3× more invasive cancers than ABUS in women with dense breasts and negative mammograms, with smaller tumor size but higher recall than ABUS.

Background

Mammography has reduced sensitivity in dense breasts, leading to missed or later-stage cancers. Supplemental imaging options include abbreviated MRI (AB-MRI), automated whole-breast ultrasound (ABUS), and contrast-enhanced mammography (CEM). Which modality best augments screening among average-risk women with dense breasts remains uncertain.

Patients

• Women aged 50–70 years in the UK NHS Breast Screening Programme with a normal full-field digital mammogram (FFDM) and BI-RADS density C or D.
• Total randomized: 9361 (AB-MRI 2318; ABUS 2240; CEM 2235; standard of care [SOC/FFDM only] 2568).
• Completed supplemental imaging and included in outcome analysis: 6305 (AB-MRI 2130; ABUS 2141; CEM 2035).
• Baseline risk distribution similar across arms; most participants were general-population risk.

Intervention

• Abbreviated MRI (AB-MRI): 10-minute contrast-enhanced protocol, double-read.
• Automated whole-breast ultrasound (ABUS): 2–3 views per breast, double-read.
• Contrast-enhanced mammography (CEM): Iodinated contrast with recombined and low-energy images, double-read.

Control

• Standard of care: Full-field digital mammography alone (FFDM).
• For comparisons with SOC, contemporaneous observational FFDM data from one center (Cambridge) in dense-breast women were used (recall 5.4%; cancer detection 8.4 per 1000).

Outcome

• Primary: Cancer detection rate (per 1000 examinations) from supplemental imaging resulting in histologically confirmed breast cancer.
• Secondary: Recall and biopsy rates; tumor characteristics (invasiveness, size, grade, nodal status, receptors); adverse events.

Study Design

• Pragmatic, multicenter randomized controlled trial (batch randomization by site/day/van) at 10 UK screening sites.
• Arms: AB-MRI, ABUS, CEM, or SOC (availability varied by site).
• Analysis: Intention-to-treat using network meta-analysis by site; primary comparison among the three active modalities; additional comparisons with SOC using observational FFDM data.
• Trial registration: NCT04097366; recruitment Oct 18, 2019–Mar 30, 2024.

Level of Evidence

Level 1 (randomized controlled trial).

Follow up period

• Interim results from the first round of supplemental imaging (typically within 6 months of screening; some exams occurred up to >6 months).
• No long-term outcomes yet; interval and next-round cancers planned over ~3 years.

Results

Primary outcome

• Cancer detection rate per 1000 examinations (any cancer):
  • AB-MRI: 17.4 (37/2130)
  • ABUS: 4.2 (9/2141)
  • CEM: 19.2 (39/2035)
• Invasive cancer detection per 1000:
  • AB-MRI: 15.0 (32/2130)
  • ABUS: 4.2 (9/2141)
  • CEM: 15.7 (32/2035)
• Number needed to screen (NNT) to detect one additional cancer vs FFDM alone (using supplemental “additional cancers” per 1000):
  • Any cancer: AB-MRI ≈ 58; ABUS ≈ 238; CEM ≈ 52
  • Invasive cancer: AB-MRI ≈ 67; ABUS ≈ 238; CEM ≈ 64

Secondary outcomes

• Recall rate: AB-MRI 9.7%; ABUS 4.0%; CEM 9.7% (FFDM-only comparator 5.4%).
• Biopsy rate: AB-MRI 4.9%; ABUS 1.5%; CEM 4.4%.
• Tumor characteristics:
  • Median invasive size: AB-MRI 10 mm (IQR 8–15); ABUS 22 mm (14–35); CEM 11 mm (7–15).
  • DCIS detected: AB-MRI 5/37 (13.5%); ABUS 0; CEM 7/39 (17.9%).
  • Nodal positivity among tested: AB-MRI 3/32; ABUS 1/9; CEM 2/32 (most were node-negative).
  • Grade (invasive): few grade 3 (3 total, all AB-MRI); majority grade 1–2 across arms.
  • Receptor status (invasive; among those tested): ER+ common across arms; HER2-enriched more frequent in AB-MRI than ABUS or CEM.
• Adverse events:
  • AB-MRI: 1 extravasation (~0.5/1000).
  • ABUS: none reported.
  • CEM: 24 iodinated contrast reactions (~11.8/1000; 17 minor, 6 moderate, 1 severe) and 3 extravasations (~1.5/1000).

Limitations

• Interim, first-round results; no long-term outcomes (interval cancers, mortality, or overdiagnosis).
• Primary analysis did not include cancers detected by FFDM alone across all sites; SOC comparisons used observational data from a single center.
• Site variability in modality availability and recall performance; learning curve effects likely.
• Some supplemental exams occurred >6 months post-screen; potential timing variability.
• Trial sized/powered for detection outcomes; not for mortality or overdiagnosis estimation.

Funding

Cancer Research UK; GE Healthcare (equipment/financial support); Bayer Healthcare (MRI contrast).

Citation

Gilbert FJ, Payne NR, Allajbeu I, et al. Comparison of supplemental breast cancer imaging techniques—interim results from the BRAID randomised controlled trial. The Lancet. Published online May 21, 2025. doi:10.1016/S0140-6736(25)00582-3