37P - CD86 expression may select patients for immune therapy in breast cancer

Date 04 May 2017
Event IMPAKT 2017
Session Welcome reception and Poster Walk
Topics Biomarkers
Breast Cancer
Translational Research
Presenter Daniel Tiezzi
Authors D.G. Tiezzi1, M. Tiezzi1, A.R. Da Silva2, J.M. Andrade1, F.F. Pimentel1
  • 1Gynecology And Obstetrics - Breast Disease Division, University of São Paulo, 14048-900 - Ribeirao Preto/BR
  • 2Pathology, University of São Paulo, 14048-900 - Ribeirao Preto/BR

Abstract

Body

Immune checkpoint blockade therapies have emerged as promising alternative for cancer treatment. Breast cancer (BC) is a heterogeneous disease and the immune therapy may be effective for some subgroups. We performed an in silico and immunohistochemistry (IHC) analyses to demonstrate there is an immune reactive subgroup.

Methods: Data from 694 patients with invasive ductal carcinoma (IDC) from public available TCGA repository were used for in silico analysis. The cytolytic score (CYT) was estimated by the geometric mean of PRF1 and GZMA genes expression. Expression data was used to infer the proportion of inflammatory cells using the CIBERSORT algorithm. PAM50 and iCluster classification was performed using the genefu and iC10 packages in R. Unsupervised hierarchal clustering was performed using expression of 22 genes related to T-cell interaction. We inferred the best number of immune clusters using the silhouette plot in cluster package. We analyzed the association of immune clusters and clinical, pathological and molecular profiles. Additionally, paraffin-embedded tissue microarray from 136 breast cancer samples were stained for PDL1, PDL2 and CD86 by IHC. The association of IHC expression and clinical and pathological data was analyzed.

Results: There is a positive correlation between CYT and expression of most genes related to T-cell interaction, especially CD86 and PDL2. Unsupervised clustering showed there are groups of IDC with high (immune reactive – C1) and low (C2) T-cell ligand expression. Most basal and iC10, iC9 and iC4ERneg tumors are in C1. The distribution of immune infiltrate is distinct across the clusters. The C1 has a higher percentage of T cytotoxic CD8+, Treg and memory T cells. It’s interesting to note the macrophage polarization. Macrophage M1 is higher in C1 and M2 is higher in C2. According to IHC, positive expression of PDL1 was 24%, PDL2 was 32% and CD86: 40%. Concomitant positive expression of all markers occurred in 10%. CD86 expression was associated to ER negative tumors (p= 0.04) and is associated to worse prognosis in multivarate analysis (HR= 2.9, p= 0.01).

Conclusion: CD86 is a potential prognostic factor in BC and identify an immune reactive subgroup of patients who may benefit from immune checkpoint blockade.

Clinical trial identification