Mammographic density and inter-observer variability of pathologic evaluation of core biopsies among women with mammographic abnormalities
1 Institute of Clinical Epidemiology, Martin-Luther-University Halle-Wittenberg, Halle (Saale), Magdeburger Str. 8, Halle (Saale), 06097, Germany
2 Cancer Registry of Norway, Postboks 5313 Majorstuen, Oslo, 0304, Norway
3 Department of Nutrition, University of Oslo, Oslo, Norway
4 Department of Preventive Medicine, University of Southern California, Los Angeles, CA, USA
5 Institute of Medical Epidemiology, Biometry and Informatics, Martin-Luther-University Halle-Wittenberg, Halle (Saale), Germany
6 Department of Radiology, Martin-Luther-University Halle-Wittenberg, Ernst-Grube-Str. 40, Halle (Saale), 06097, Germany
7 Institute of Pathology, Martin-Luther-University Halle-Wittenberg, Magdeburger Str. 14, Halle (Saale), 06097, Germany
8 Albertinen-Pathology Hamburg, Fangdieckstr. 75 a, Hamburg, 22547, Germany
9 Department of Gynaecology, Martin-Luther-University Halle-Wittenberg, Ernst-Grube-Str. 40, Halle (Saale), 06097, Germany
10 Reference Center for Gynaeco- and Mammapathology, Fangdieckstr. 75 a, Hamburg, 22547, Germany
BMC Cancer 2012, 12:554 doi:10.1186/1471-2407-12-554Published: 24 November 2012
As high percentage of mammographic densities complicates the assessment of imaging findings, mammographic density may influence the histopathological evaluation of core-biopsies of the breast. We measured the influence of mammographic density on the inter-observer variability of histopathological findings of breast biopsies.
Histological slides of 695 women who underwent core biopsies of the breast at University of Halle between 2006 and 2008 were evaluated in a blinded fashion by two pathologists using the five levels of the B-categorization scheme (B1-B5). To quantify mammographic density, we used a computer-based threshold method (Madena). We calculated observed and chance-corrected agreements (weighted kappa) and 95% confidence intervals (95% CI) according to four categories of mammographic density (<10%, 10<25%, 25<50%, ≥50%).
The weighted kappa decreased monotonically from 89.6% (95% CI: 85.8%, 93.3%) among women with less than 10% of mammographic density to 80.4% (95% CI: 69.9%, 90.9%) for women with more than 50% of mammographic density, respectively. Results of a kappa regression analysis showed that agreement of pathologists on clinically relevant categories (B1-B2 versus B3-B5) decreased with mammographic density.
Mammographic density is a relevant modifier of the agreement between pathologists who assess breast biopsies using the B-categorization scheme. The influence of mammographic density on the inter-observer variability can be explained to some extent by varying prevalences of histological entities across B categories that have typically different inter-observer agreement. Women with high mammographic density are at higher risk of inter-observer variability compared to women with low mammographic density and should possibly undergo a second pathology review.