Eight studies (n=4,331) assessed the screening performance of MRI (evidence levels 2b to 4); the same studies appeared to also assess the screening performance of mammography. A subset of five studies (n=3,389) assessed the performance of screening with MRI or mammography.
MRI alone: The pooled estimate of cancer detection rate was 0.020 (95% CI 0.011 to 0.028), the pooled estimate of sensitivity was 0.966 (95% CI 0.946 to 0.986) and the pooled estimate of the false positive rate was 0.095 (95% CI 0.037 to 0.153).
Mammography alone: The pooled estimate of cancer detection rate was 0.007 (95% CI 0.004 to 0.009), the pooled estimate of sensitivity was 0.375 (95% CI 0.298 to 0.45) and the pooled estimate of the false positive rate was 0.041 (95% CI 0.024 to 0.057).
MRI and mammography combined: The pooled estimate of cancer detection rate was 0.019 (95% CI 0.010 to 0.029), the pooled estimate of sensitivity was 0.944 (95% CI 0.896 to 0.992) and the pooled estimate of the false positive rate was 0.138 (95% CI 0.039 to 0.237).
Across the whole population, the performance characteristics of MRI alone were significantly better than those for mammography alone, and did not differ significantly from those for MRI and mammography combined. The performance characteristics of each imaging technique did not differ significantly when pooled studies were restricted to those of evidence level 2 and above, conducted in women with BRCA1/2 mutations (three studies). The results of heterogeneity assessment were only reported for the sensitivity analysis; heterogeneity was high in all cases.