The accuracy of magnetic resonance imaging in predicting the size of pure ductal carcinoma in situ: a systematic review and meta-analysis.

Ductal carcinoma in situ (DCIS) is a putative precursor of invasive breast cancer and MRI is considered the most sensitive imaging technique for its detection. This study aims to evaluate the accuracy of MRI measuring the pure DCIS size, against pathology, to better understand the role of MRI in the management of this intraductal neoplasm.Potentially eligible studies in MEDLINE, Embase and Google Scholar, up to January 2021 were considered, and a systematic review and meta-analysis according to the published protocol (Prospero-CRD42021232228) was performed. Outcomes of mean differences and accuracy rates were analysed using IBM® SPSS® v26 and random-effect models in platform R v3.3.Twenty-two cross-sectional studies were selected and 15 proceeded to meta-analysis. MRI accurately predicted 55% of the tumours' sizes and, according to Bland-Altman plots, concordance between MRI and pathology was greater for smaller tumours. In the meta-analysis, difference of the means between MRI and pathology was 3.85 mm (CI 95% [-0.92;8.60]) with considerable heterogeneity (I2 = 96.7%). Subgroup analysis showed similar results for sizes between different MRI fields, temporal resolution, slice thickness and acquisition times, but lower heterogeneity in studies using 3-T MRI (I2 = 57.2%). Results were concordant with low risk of bias studies (2.46, CI 95% [0.57-4.36]), without heterogeneity (I2 = 0%).Therefore, MRI is shown to be an accurate method in pure DCIS size assessment. Once the best MRI protocol is established, evaluation of the impact of pure DCIS size in predicting treatment outcomes will contribute to clarifying current issues related to intraductal breast carcinoma.