Abstract P1-07-07: Gene expression signatures and immunohistochemical subtypes add prognostic value to each other

Abstract

Abstract Background: We have previously demonstrated that gene expression signatures and Ki67 stratify the same breast tumour into opposing good/poor prognosis groups in approximately 20% of cases. Given this, we hypothesized that the combination of a clinically relevant gene signature and IHC markers may provide more prognostic information than either classifier alone. We tested this hypothesis in a large independent cohort of Swedish breast cancer patients with long-term follow-up data. Methods: We assessed Ki67, ER, PR, HER2 and the research versions of the Genomic Grade Index (GGI), Mammaprint, cell-cycle score (CCS), Recurrence Score (RS) and PAM50 gene expression classifiers on matching TMA and microarray data in a Swedish breast cancer cohort of 623 patients. Change in likelihood-ratio (Δ LR-χ2) was used to first determine the additional prognostic information provided by gene expression signatures beyond that provided by 1) Ki67 alone and 2) Ki67 plus ER, PR and HER2, grouped to form the IHC molecular subtypes. Secondly and conversely, we then determined the additional prognostic information provided by Ki67/IHC subtypes beyond gene expression signatures. Results: Representative images from Ki67/gene signature contrast groups show tumours with high levels of Ki67 expression that are classified as good prognosis by gene signatures and conversely, tumours with low Ki67 that are classified into poor prognosis groups by gene signatures. In all patients (n=623), the majority of signatures provided statistically significant information beyond that of Ki67 alone, however only RS and PAM50 remained significant in the presence of the IHC subtypes (Δ LR-χ2 RS= 11.7 and PAM50 = 15.4; P = 0.002 and 0.004, respectively). Conversely, IHC subtypes added prognostic information beyond gene signatures whilst Ki67 alone did not, a notable exception to this was PAM50. Conclusions: In general, a combination of the IHC subtypes with gene signatures provides more prognostic information than either classifier alone when considering all breast cancer patients. Subsequent analyses will focus on patient subgroups including ER positive, node positive and ER positive, node negative groups, along with validation of our work in a second dataset of 253 patients. Change in likelhood ratio with the addition of gene expression signatures to Ki67/IHC subgroups and vice-versa All Patients All PatientsSig. added to Ki67:Sig. Δ LRχ2P-valueSig. added to IHC subtypesSig. Δ LRχ2P-valueGGI6.00.014GGI2.50.108Mammaprint6.30.011Mammaprint1.10.279RS20.8< 0.001RS11.70.002CCS1.70.409CCS2.00.360PAM5025.0< 0.001PAM5015.40.004 Ki67 added to sig.:Ki67 Δ LRχ2P-valueIHC added to sig.:IHC Δ LRχ2P-valueGGI1.60.205GGI14.90.001Mammaprint1.60.199Mammaprint15.30.001RS0.50.477RS12.60.005CCS4.10.041CCS16.10.001PAM502.30.13PAM506.10.107Sig.: Gene expression signature; GGI: Genomic grade index; RS: Recurrence score; CCS: Cell cycle score. Citation Format: Lundberg A, Lindström LS, Falato C, Carlson JW, Foukakis T, Czene K, Bergh J, Tobin NP. Gene expression signatures and immunohistochemical subtypes add prognostic value to each other [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P1-07-07.