Abstract P4-06-03: Assays on core biopsies and surgically resected tumors may result in different subtyping of the invasive breast cancer from the same patient

Background Core biopsies (CBs) are often used for biomarker expression assays to determine the treatment regimen. However, a number of other clinically important analyses (e.g. OncoType Dx), are performed on surgically resected tumors (SRTs). A previous study has shown that biomarkers ER, PR, and Ki67 expressed higher in CBs than in SRTs. Here we analyze how this difference impacts the subtyping of ER+ breast tumors.Methods Female patients enrolled in the Clinical Breast Care Project (CBCP) from a civilian site were selected for this study, where expression of ER, PR, HER2, and Ki67 were assayed by IHC in a reference lab on CBs; the same 4 assays were performed on SRTs by a CBCP central lab. Both labs are CLIA-certified. Patients treated with neoadjuvant chemotherapy and those with multiple tumors were excluded. 167 cases were identified for this study to compare assays performed on CBs and SRTs from the same patients. ER and PR were positive if >1% nuclear staining, HER2 was negative if IHC = 0 or 1+, positive if IHC = 3+, and for IHC = 2+ FISH was used for the final call. Ki67 was positive if > = 15% nuclear staining. LA was ER+/HER2-/Ki67-, LB1 was ER+/HER2-/Ki67+, and LB2 was ER+/HER2+. For histologic grades, only readings from the central lab on SRTs were used. Statistical analyses were performed using SAS.Results This analysis confirmed that Ki67, ER, and PR showed higher percent nuclear staining in CBs than in SRTs from the same patients. The difference for Ki67 was more striking and unidirectional. ER and PR cases clustered at the upper percent levels. Histograms with a bin-width of 15% show a peak at 15% for Ki67 difference between CBs and SRTs, whereas the peaks for ER and PR differences were at 0%. McNemar's (or Exact McNemar’s) test showed significant differences between the binary status calls for Ki67 (p = 3.2E-15) and ER (p = 0.012), but not for PR (p = 0.65). Assays on CBs and SRTs resulted in different subtype calls for the cases (Table 1). Grade distributions were different between LA and LB (p<0.001 for both CB- and SRT-based subtypes, Chi-Square or Fisher's Exact test), but not so between LB1 and LB2 (p = 0.23 for CB, 0.31 for SRT). However, SRT-based LB1 cases concentrate more on higher grades compared to CB-based cases (p = 0.048).Table 1. ER+ subtypes based on IHC assays (from CBs and SRTs) and corresponding grades (from SRTs) CBSRTSubtypeG1G2G3G1G2G3LA2126034518LB11435342820LB2036032Discussion On IHC assays, Ki67 expression is strikingly higher in CBs than in SRTs, and ER expression is also higher in CBs than in SRTs. This directly resulted in more LB than LA subtypes based on CBs. SRT-based LB1 cases concentrate more on higher grades compared to CB-based cases, which is more consistent with the observation that LB subtypes have worse outcomes. A limitation of this study is that technical differences between the labs may contribute to the observed differences between CBs and SRTs. Further studies need to be performed to determine whether SRT should also be assayed in addition to CB for treatment regimen decision-making.The views expressed in this abstract are those of the authors and do not reflect the official policy of the Department of Defense, or US Government.Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P4-06-03.

Duke Scholars

Author Terry Hyslop Biostatistics & Bioinformatics, Division of Translational Bi ...