Abstract PO4-25-10: Towards personalized medicine for DCIS - the role of hormone receptors, HER2, and Ki67 status in high-grade DCIS

Abstract Objective: This study aimed to investigate the molecular details in breast ductal carcinoma in situ (DCIS). Expression status of estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor receptor 2 (HER2), and Ki67 were investigated among luminal A (LumA), luminal B HER2-negative (LumB HER2-), luminal B HER2-positive (LumB HER2+), HER2-enriched, and triple-negative (TPN), subtypes of DCIS. Materials & Methods: The study comprised formalin-fixed paraffin-embedded (FFPE) specimens of 357 DCIS grade 3 cases diagnosed between 1996 – 2018. Routine diagnostic immunohistochemical (IHC) staining were performed. DCIS cases were classified as LumA, LumB HER2-, LumB HER2+, HER2-enriched or TPN, according to the 2013 St. Gallen guidelines, which is used for molecular subtyping of invasive breast carcinoma (IBC). Each subtype was sorted into three subcategories: “pure” meaning those without an invasive component; “w/invasive” meaning those with an invasive component; and “all” meaning the entire group of the given subtype. Furthermore, ER and PR-receptor expression was registered for LumA, LumB HER2- and LumB HER2+ cases as intervals. The distribution of Ki67 was analyzed within each subtype. For some analyses, we combined the LumA and LumB HER2- subtypes into one entire group. The inter-observer variability of Ki67 was calculated by setting a cut-off value of 20% (2013 St. Gallen). A cohort (n = 47) of DCIS cases with a median Ki67 value of 18% was selected. Ki67 was calculated by analyzing 200 DCIS cells in two separate hotspots. Cohen's Kappa coefficient was calculated based on these data. HER2 (IHC) was scored based on ASCO/CAP guidelines established for routine diagnostic work-up for IBC. 16 equivocal cases (2+) were further investigated using dual SISH. Results: 98% of “all” cases of the LumA subtype showed an ER ≥ 50%. PR expression ≥ 50% was found in 91% of cases in this subtype. The incidence of ER-receptor at a cut-off ≥ 50% in the LumA subtype was significantly higher than that in the LumB HER2- and LumB HER2+ subtypes (p < 0.0001, Chi-square). In contrast, there was a statistically significant reduction in ER-receptor expression at a cut-off < 10% in LumA compared to the latter subtypes. The proportion of cases with PR-receptors with a cut-off of < 20% showed significant differences between LumA, LumB HER2- and LumB HER2+ subtypes (1.6%; 47% and 37%, respectively). There was also a significantly higher proportion of PR-receptor ≥ 50% cases among the LumA subtype. We found a significant association between PR < 20% and HER2 (3+) in luminal subtypes (p < 0.0004, Fisher's exact). There was no significant difference in ER/PR expression in “pure” cases of luminal subtypes of DCIS versus those with an invasive component (p = 0.1831, Chi-square). The Ki67 in the entire population (n = 357) varied from < 1% to > 80%. The mean and median were around 20% in those subtypes whose classifications were not depended on Ki67. There was a significant difference in the distribution of Ki67 when cases of LumA “all” and those of LumB HER2- “all” were combined as one entire group and were compared to LumB HER2+ “all” and HER2-enriched “all” (p-adjusted < 0.0001, Kruskal Wallis). The assessment of Ki67 among different observers showed a Cohen's kappa coefficient of 0.29 – 0.31 (fair agreement). We compared the HER2 (IHC) 0, 1+ and 2+ score among LumA and LumB HER2- subtypes and did not find a statically significant difference, when the “pure” and “w/invasive” were compared (p = 0.603, Chi-square). Conclusions: Ki67 was highly variable in DCIS grade 3. Inter-observer agreement was (as expected) suboptimal, and the cut-off at 20% defined by the 2013 St Gallen guidelines for IBC is not reliable for the distinction of LumA and LumB HER2 subtypes of DCIS. The LumB subtype of DCIS is heterogeneous with considerable variability among the four IHC markers used in the present study. A low PR is strongly associated with HER2 (3+), in luminal subtypes (p < 0.0004, Fisher's exact). Citation Format: Hossein Schandiz, Lorant Farkas, Daehoon Park, Solveig Norheim Andersen, Jürgen Geisler, Torill Sauer. Towards personalized medicine for DCIS - the role of hormone receptors, HER2, and Ki67 status in high-grade DCIS [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO4-25-10.