Abstract 43P
Background
Triple-negative breast cancer's (TNBC) adverse prognosis necessitates enhanced therapeutic strategies. The recognized molecular classifications, illustrate its heterogeneity yet omit spatial analysis and remain non-integrative in clinical protocols. This study intends to assess TNBC biomarker expression through immunohistochemistry (IHC) and correlate it with spatial gene expression profiles derived from spatial transcriptomics (ST).
Methods
We examined a retrospective cohort of patients (pts) at Institut Jules Bordet, Belgium, undergoing primary surgery for early TNBC. ST (Visium® Spatial Gene Expression, 10X Genomics) on fresh frozen samples yielded spatial gene expression from tumor, stroma, and structures. Duplex IHC for Trop-2/AR and HER2/Ki-67 was done on serial sections, scored by H-score (Trop-2), Allred (AR), and guidelines (Ki67/HER2), adding ultra-low and null HER2. Analysis included descriptive stats, Spearman's for IHC-ST correlation, Mann-Whitney for gene expression, and Kaplan-Meier with log-rank for survival.
Results
Among the 92 pts included, Trop-2 IHC expression was high in 67% of pts, correlated with AR by IHC (r=0.3, p=0.004) and anti-correlated with Ki67 (r=-0.34). IHC expression correlated well with tumor-derived gene expression (r=0.54) and was inversely associated with prognosis (p=0.021). AR expression was scored both in tumor and stromal cells and was correlated (r=0.46) and also correlated with the respective gene expression (r=0.68). Stromal but not tumor AR was correlated with TILs (r=0.26, p=0.014), while only tumor AR correlated weakly with Ki67 (r=-0.21, p=0.04). Neither of the two was associated with prognosis. Importantly, using AR and Ki67 expression levels by IHC, we could construct a classifier that accurately identified LAR tumors, with a PPV of 93% and an NPV of 96%. HER2 by IHC was scored 2+, 1+, ultra-low, and null in 4.3%, 27.2%, 30.4%, and 38% of pts, respectively. Importantly, ERBB2 tumor gene expression in HER2 ultra-low tumors was significantly different compared to HER2 null tumors (p<0.0001), but similar to HER2 1+ tumors (p=0.49).
Conclusions
Converting higher into lower resolution data might be feasible and pave the path for tailored therapies in TNBC.
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.