Borderline Ductal Epithelial Lesions: Interobserver Reproducibility in 43 Cases with Original Diagnosis of Atypical Ductal Hyperplasia Bordering on Ductal Carcinoma In-Situ
Gary Tozbikian, Edi Brogi, Jeffrey Catalano, Sujata Patil, Kimberly J Van Zee, Maryam Kadivar, Christina E Vallejo, Yong H Wen. Memorial Sloan-Kettering Cancer Center, New York, NY; Rasool-Akram Hospital, Tehran University of Medical Sciences, Tehran, Islamic Republic of Iran
Background: The distinction between atypical ductal hyperplasia (ADH) and ductal carcinoma in situ (DCIS) has substantial clinical impact. In most cases use of established diagnostic criteria (Page DL, Rogers LW. Hum Path 1992) allows one to solve this differential, but occasionally pathologists render a diagnosis (dx) of ADH bordering on DCIS (ADH/DCIS). We studied diagnostic interobserver reproducibility in these borderline lesions.
Design: We identified patients (pts) who had breast surgical excision (EXC) at our center between 1997 and 2008 and received dx of ADH/DCIS. Patient age and specimen type (core biopsy (CBX) or EXC) were noted. A study pathologist identified the diagnostic slides. Four pathologists (aware of the index dx) independently scored each case as benign, ADH, ADH/DCIS or DCIS. They assessed histologic pattern, lesion size (≤2 or >2 mm), number of involved (#) ducts (≤ 2 or >2), and nuclear grade (NG) (1 or >1). Interobserver variance was calculated by k-statistic. Complete scores (available for 3 reviewers) were averaged into a consensus score (CDx) and correlated with pathologic variables by Fisher test.
Results: Slides from 43 pts with EXC and dx of ADH/DCIS were available for study. Pt median age was 52+11 years. Indications for EXC were Ca2+ (25), mass (10) and unknown (8). 28 pts had prior CBX. ADH/DCIS dx pertained to 12 CBXs, 30 EXCs, and to both for 1 pt. The lesion median size was 2.4 mm (95%CI: 1.8-2.8). ADH/DCIS was the most frequent CDx (25/43; 58%), followed by ADH (14/43; 33%) and DCIS (4/43; 9%). CDx of DCIS correlated significantly with # ducts, but not with size, NG, or histologic pattern. [Table 1] There was complete (4/4) agreement in 14% of cases (4 ADH, 1 ADH/DCIS, 1 DCIS) and near agreement (3/4) in 35% (9 ADH, 4 ADH/DCIS, 2 DCIS). Interobserver agreement was slight for diagnosis (k= 0.11), # ducts (k =0.09), NG (k= 0.13); and moderate for size (k= 0.4).
|ADH (n=14)||ADH/DCIS (n=25)||DCIS (n=4)||p value||k|
|# Ducts||≤ 2||12||19||1||0.045||0.09|
|Size (mm)||≤ 2||9||16||2||0.6||0.40|