Technical aspects of breast MRI—do they affect outcomes?

Abstract

In a systematic review of breast MRI for assessing ipsilateral breast cancer to detect additional lesions, technical details were extracted from publications to assess their effect on diagnostic performance. Where technical parameters were complete, we examined their effect on summary ROC models, and the TP:FP ratio and PPV, using random-effects logistic regression. A total of 2,801 breasts in 19 publications underwent statistical analysis for year of study, slice thickness, and repetitions after contrast-medium injection. None were associated with TP/FP ratio. Summary ROC analysis provided weak evidence (P = 0.09) of an association between diagnostic performance and time period, however no trend over time. Tesla strength was reported in 2,801 cases. Other key information was omitted including whether both breasts were examined for 1683 (60%), position of the patient in 1,375 (49%), and imaging planes used in 688 (25%). Contrast agent and dose were reported for 2,646 (95%) breasts. Reporting technique was inconsistently reported. Single radiology reports were found in 1,637 (58%) cases, double in 347 (12.4%), and in 960 (34%) knowledge of mammography or ultrasound findings was not stated. Slice thickness, number of sequences after contrast medium, and year of study did not show significant performance differences. Other technical information was deficient. There is an urgent need to improve the quality of reporting of breast MRI studies.