How it started?
When I joined Prof. Vessela Kristensen group in 2016, I started working with Dr. Thomas Fleischer on integrating DNA methylation and gene expression data. It was such fun working on millions of correlations between CpG methylation and gene expression that it led to a great publication also in Nature Communications (1). In this previous paper, in addition to the discovery of a strong regulatory gene network under the control of estrogen receptor, we also found another, equally strong network, corresponding to immune signaling. This and other studies from our lab (2,3) lead to the hereby described project which aimed at better understanding the role of immune infiltration in breast cancer.
Hypothesis and Challenges
Motivated by recent studies showing how the immune microenvironment of a tumor can be dissected using transcriptomic data from the bulk tumor, we aimed at using gene expression to understand how immune infiltration may become clinically relevant in breast cancer. The two main challenges we faced were:
- Only the expression of few genes can reliably be used to infer immune context.
- Many before us have used gene expression data to classify breast cancer into clinically relevant subtypes.
However, no one had described breast cancer subtypes with biological properties reflecting the immune context of the tumor.
1. We characterize breast cancer immune clusters/subtypes and demonstrate their independent prognostic value. Subtyping breast cancer patients by PAM50 and by immune clusters allows to pinpoint inside each PAM50 subtype, a subgroup of patients with a worse prognosis.
2. Immune clusters/subtypes are associated with response to neoadjuvant chemotherapy and help to pinpoint which patients may benefit from new treatment strategies.
3. The immune subtype/cluster with the worse prognosis is composed of pro-tumorigenic (M2 type of Macrophages) and inactive immune cells (resting Mast cells). This is in contrast to the two other clusters, which are immune hot (high levels of active immune cells) and immune cold (few/no infiltrating immune cells).
4. Through phenotypical analyses of the immune clusters, we find that proliferation and epithelial mesenchymal transition are mutually exclusive in breast cancer.
We demonstrate the clinical relevance of the immune clusters with reference to further understanding the tumor microenvironment and harvest its potential in improving breast cancer management.
We provide an online service for researchers to subtype their favorite cohort by immune clusters. Our hope is that many will use it and will be in contact with us to report on performance. We will continue improving the current method (described in the publication) to subtype immune clusters; feedbacks from others using it will help us.
We believe that this study will contribute to the knowledge needed for giving women personalized diagnosis and treatment options. The number and complexity of the immune cell types found in the tumor microenvironment is enormous. Our current description of the composition of the immune clusters is based on in silico predictions. Using new powerful techniques such as in situ RNA hybridization and single cell RNA-seq will bring us one step closer to understanding how breast tumors shape their microenvironment and vice versa how immune and stromal cells influence tumor evolution and survival.
Dr. Xavier Tekpli: firstname.lastname@example.org
Prof. Vessela Kristensen: email@example.com
1. Fleischer T, Tekpli X. et al. DNA methylation at enhancers identifies distinct breast cancer lineages. Nat Commun 8, 1379 (2017).
2. Kristensen V.N. et al. Integrated molecular profiles of invasive breast tumors and ductal carcinoma in situ (DCIS) reveal differential vascular and interleukin signaling. Proc Natl Acad Sci U S A 109, 2802-7 (2012).
3. Quigley D. et al. Lymphocyte Invasion in IC10/Basal-Like Breast Tumors Is Associated with Wild-Type TP53. Mol Cancer Res 13, 493-501 (2015).