Morphological variants of lobular carcinoma in situ (LCIS) include classical (CLCIS), pleomorphic (PLCIS) and florid type (FLCIS). Treatment guidelines suggest managing PLCIS and FLCIS like ductal carcinoma in situ (DCIS); therefore accurate identification of LCIS subtypes is critical. However, the significance of separating PLCIS from FLCIS is not clear. Also, interobserver agreement in identifying LCIS subtypes, using contemporary criteria, is not known. We aimed to evaluate interobserver agreement amongst breast pathologists in diagnosing LCIS subtypes and use the agreement data to justify LCIS classification for management purposes. Six breast pathologists independently reviewed 50 hematoxylin and eosin-stained slides comprised of a mix of LCIS subtypes. After reviewing published criteria, participants diagnosed PLCIS, CLCIS and apocrine change in a marked region of interest and FLCIS based on entire section. PLCIS was identified in 8 to 37 slides with overall moderate agreement (Fleiss' κ = 0.565) and pairwise κ (Cohen's) ranging from -.008 to 0.492. FLCIS was diagnosed in 15-26 slides with overall substantial agreement (Fleiss' κ = 0.687) and pairwise κ ranging from -.068 to 0.706. Both FLCIS and PLCIS coexisted in 45% of slides with consensus on non-classical LCIS. Comedo-type necrosis (odds ratio = 5.5) and apoptosis (odds ratio = 1.8) predicted FLCIS. We found moderate and substantial agreement in diagnosing PLCIS and FLCIS respectively. Objective histological features linked with aggressive behavior were more frequent with FLCIS. PLCIS and FLCIS patterns frequently coexist, contain similar molecular aberrations, and are managed similarly (like DCIS); therefore, combining FLCIS and PLCIS into one category (non-classical LCIS) should be considered.
Keywords: Agreement; Apocrine; Breast cancer; Florid; Interobserver variability; Lobular carcinoma; Pleomorphic.
Published by Elsevier Inc.