Five studies provided data on accuracy (n=2,059), of which four (n=1,553) also reported on false-positive recall and benign biopsy rates.
Accuracy (5 studies).
The sensitivity of MRI combined with conventional tests ranged from 86 to 100%, and the specificity ranged from 77 to 96%. The sensitivity of mammography alone ranged from 25 to 59%, and from 49 to 67% for mammography combined with ultrasound with or without CBE.
Three studies assessed MRI added to mammography alone: the pooled sensitivity was 94% (95% CI: 86, 98) and the pooled incremental sensitivity was 58% (95% CI: 47, 70). There was no evidence of heterogeneity (p=0.84).
Three studies assessed MRI added to mammography plus ultrasound: the pooled incremental sensitivity was 44% (95% CI: 27, 61).
Two studies assessed MRI added to mammography plus ultrasound plus CBE: the respective incremental sensitivities were 31 and 33%.
Additional cancer yield (4 studies).
The additional cancer yield in women with negative findings based on conventional tests ranged from 10 to 24 additional cancers detected per 1,000 screening rounds. Four studies reported that
mammography detected cancers not detected by MRI, and one also reported that ultrasound detected cancers not detected by MRI or mammography. CBE detected cancers that were not found by mammography or ultrasound, but all cancers found by CBE were also detected by MRI.
Test recall rates (3 studies).
The risk of being recalled for further investigation where cancer was subsequently excluded was 3 to 5 times greater with MRI combined with mammography than with mammography alone; this is
equivalent to 71 to 74 additional false-positive recalls per 1,000 screening rounds. The relative risk of undergoing a benign percutaneous biopsy as a result of the addition of MRI ranged
from 1.22 to 9.50; this is equivalent to 7 to 46 additional benign biopsies per 1,000 screening rounds.
Stage shift in cancer detection.
The size of the tumours detected with the addition of MRI was similar to those detected without MRI.