28P The Ki67 proliferation index in subtypes of human DCIS grade III: A biomarker with potential impact on clinical decision making?

This study investigated the Ki67 proliferation index in human high-grade DCIS with respect to the Luminal (Lum) A, LumB HER2ˉ, LumB HER2+, HER2-enriched, and triple-negative (TPN) subtypes. Additionally, we attempted to determine a more reliable Ki67 cutoff value. The study included formalin-fixed paraffin-embedded specimens of 357 high-grade DCIS cases diagnosed between 1996-2018. Routine diagnostic immunohistochemical staining was used. DCIS cases were classified as LumA, LumB HER2ˉ, LumB HER2+, HER2-enriched or TPN, according to the 2013 St. Gallen guidelines. Each subtype was sorted into: "pure": without an invasive component; "w/invasive": with an invasive component; and "all": the entire group of the given subtype. The distribution of Ki67 was studied within each subtype. By applying Cohen's kappa coefficient and a 20% cutoff, the interobserver diagnostic agreement between counting by pathologist 1 (CP1) and CP2 on the Ki67 assessment was determined. We looked at the relationship with HER2 (score 3+) and the distribution of low PR (<20%) together with high Ki67 (≥20%) among LumB HER2ˉ, LumB HER2+, HER2-enriched, and TPN subtypes. In subtypes in which their classification was not dependent on the Ki67 we identified significant differences among the selected groups (p<0.0001). The median Ki67 was significantly higher in cases with invasive components than in those without invasive components; 25 and 19, respectively (p=0.0351). A combination of low PR level and high Ki67 expression was significantly associated with HER2 overexpression (score 3+) (p=0.0107). The interobserver diagnostic agreement for Ki67 between CP1 and CP2 was moderate (ƙ=0.58). Ki67 was considerably variable in high-grade DCIS. Low PR combined with high Ki67 was associated with HER2-enriched subtype and HER2 overexpression (score 3+). The interobserver agreement was moderate. Our results indicate that the use of a Ki67 cutoff of 20% as defined by the 2013 St. Gallen are probably not entirely reliable for distinguishing luminal subtypes of high-grade DCIS. However, Ki67 may have a role in identifying those DCIS with invasive potential, paving the way for a new clinical decision making for treatment of DCIS in the future.