SAN ANTONIO – Robustness of dichotomous assessment is important when evaluating stromal features of ductal carcinoma in situ (DCIS), according to a study presented at the 2016 San Antonio Breast Cancer Symposium (SABCS).1

“DCIS is regarded as a nonobligate, preinvasive precursor of invasive ductal carcinoma,” said Mieke Van Bockstal, MD, Phd, Department of Pathology, Ghent University Hospital, Ghent, Belgium. “Adequate prognostic markers require robustness of assessment.”

Investigators previously demonstrated that DCIS with a predominantly myxoid periductal stromal architecture is associated with an increased risk of both overall and invasive disease recurrence. Therefore, researchers sought to determine a cut-off for the assessment of myxoid stroma and stromal inflammation in DCIS, based on interobserver variability.

Continue Reading

For the study, researchers assessed hematoxylin/eosin stained sections from a cohort of 435 patients with pure DCIS diagnosed between 2000 and 2009 who underwent either mastectomy or breast-conserving surgery.

Two pathologists independently evaluated nuclear grade, intraductal calcifications, necrosis, DCIS architecture, stromal architecture, and stromal inflammation.

“Stromal architecture was classified into 4 categories: 0%, 1-33%, 33-66%, or greater than 66% of ducts surrounded by myxoid stroma,” explained Dr Van Bockstal. “Similarly, stromal inflammation was semi-quantitatively classified into 4 categories: designated absent, mild, moderate, or extensive stromal inflammation.”

Researchers also used kappa values, a measure of inter-rater agreement, to assess interobserved variability.

Results showed that the kappa value for scoring myxoid stromal architecture was highest by dichotimization with a cut-off at 33% (κ 0.523), compared with κ 0.478 and κ 0.505 when using 1% and 55% as cut-off levels, respectively.

“Concordance among both observers amounted 81.4% when scores were dichotomized with a cut-off at 33%, compared with 74.5% and 90.8% concordance levels when using 1% and 66% myxoid stroma as a cut-off,” said Dr Van Bockstal. “The 33% cut-off was considered most robust.”

For stromal inflammation, the highest value was obtained by dichotimization as ‘absent to mild’ vs ‘moderate to extensive’ inflammation (κ 0.738). Again, the 33% cut-off was determined to be the most robust.

“The dichotomous assessment of stromal features in DCIS resulted in similar or even higher kappa values compared to the dichotomous scoring of other histopathologic characteristics,” concluded Dr Van Bockstal. “This study demonstrates the robustness of dichotomous assessment of stromal architecture and stromal inflammation.”


1. Van Bockstal M, Lambein K, Smeets A, et al. Assessment of stromal characteristics in ductal carcinoma in situ of the breast: An inter-observer variability study. Poster presented at: 2016 San Antonio Breast Cancer Symposium; December 6-10, 2016; San Antonio, TX.