Karen Drukker PhD, Presenter: Royalties, Hologic, Inc

Maryellen L. Giger PhD, Abstract Co-Author: Stockholder, Hologic, Inc Shareholder, Quantitative Insights, Inc Royalties, Hologic, Inc Royalties, General Electric Company Royalties, MEDIAN Technologies Royalties, Riverain Technologies, LLC Royalties, Mitsubishi Corporation Royalties, Toshiba Corporation Researcher, Koninklijke Philips Electronics NV Researcher, U-Systems, Inc

Evaluate variability in the clinical assessment of breast images, and its dependence on radiologists’ professional and practice characteristics, in a retrospective reader study combining X-ray mammography (XRM) and 3D automated breast ultrasound (ABUS) for breast cancer detection in women with dense breasts.

The study involved 17 breast radiologists of which 7 came from academic radiology practices, 6 from private practice, and 4 from community clinics.  A sequential study design was employed with readers first interpreting ‘XRM alone’ followed by an interpretation of the combined ‘XRM+ABUS’, with each interpretation including a forced BI-RADs scale and a likelihood that the woman had breast cancer. The analysis included 164 asymptomatic patients, including 31 breast cancer patients, with dense breasts and a negative screening XRM. Of interest were inter-reader variability in scoring for ‘XRM alone’, ‘XRM+ABUS’, and the dependence on reader experience, fellowship training, and type of practice. Performance analysis included Receiver Operating Characteristic (ROC), percentile, Kappa statistics, correlative, and Bland-Altman analyses. The statistical significance of the impact of consecutive reads was assessed for the kappa statistics using bootstrapping.

The median change in area under the ROC curve after ABUS interpretation was 0.12 (range 0.04—0.19). Reader agreement was fair with the median inter-reader kappa being 0.26 (0.05—0.48) for ‘XRM alone’ and 0.34 (0.11—0.55) for ‘XRM+ABUS’ (95% confidence interval for the difference in kappa [0.06;0.11]). The only factor that appeared to have a substantial effect on reader performance was the type of clinical radiology practice, with the increase in area under the ROC curve the largest for the 3 radiologists from academic practices, with changes of 0.18, 0.19, and 0.19 respectively.

A modest, but statistically significant, increase in inter-reader agreement was observed after interpretation of ABUS, while radiologists from academic practice seemed to benefit the most from ABUS interpretation.

Understanding reader variability and factors such as training and clinical practice will yield informed decisions on the use of multimodality imaging in breast cancer screening.

Drukker, K, Giger, M, Impact of Radiologists’ Professional and Practice Characteristics on Breast Cancer Detection in Women with Dense Breasts; A Reader Study Combining Mammography and Automated Breast Ultrasound.  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13012678.html