Integrated mathematical and experimental analysis of immune system signaling in breast cancer

Overview

Project Title

Integrated mathematical and experimental analysis of immune system signaling in breast cancer

Project Website

https://www.cityofhope.org/people/lee-peter/
https://www.cityofhope.org/people/rockne-russell
https://www.cityofhope.org/people/rodin-andrei

Project Summary

Our project seeks, via an integrated experimental–computational approach, to unravel the complexity of immune signaling networks in healthy individuals and patients with estrogen receptor-positive breast cancer (ER+ BC). Our focus is on the identification and characterization, at a systems level, of the dynamical flow of information through immune signaling networks in healthy individuals and ER+ BC patients, leading to a deeper understanding of how immune signaling defects at diagnosis reflect patients’ immune responses and eventuate into different clinical outcomes. The systems biology and dynamic modeling tools and methods that are being developed are, of course, generalizable within a broader immune-oncological context.

Approximately 40% of patients with estrogen receptor-positive breast cancer (ER+ BC, of any stage) harbor immune signaling defects in their blood at diagnosis. Importantly, these defects can predict cancer relapse years later and thus may be developed into a prognostic blood test. Furthermore, rational modulation of the immune system to correct signaling defects offers a novel approach to improving outcomes for ER+ BC patients. However, the underlying mechanisms and downstream consequences of these immune signaling defects remain poorly understood at a systems (network) level. There is also a gap in our understanding of how multiple signaling elements of the immune system and tumor cells act in concert. To integrate key signaling pathways in immune responses to cancer and model their interactions as a system—a complexity compounded by the intrinsic diversity of the immune system—we are pursuing a multidisciplinary and systems biology computational research strategy.

Specifically, our approach is to experimentally capture a rich data set spanning multiple heterogeneous biological inputs (molecular, genetic, and cellular data), then to filter and integrate these data using Bayesian Networks (BNs) to build probabilistic relationships between the disease state and various nodes of interest within the BNs reflecting key multimodal biological variables (e.g. gene expression, protein modifications). These are also being augmented by the dynamical modeling approaches. Our early results revealed major cytokine signaling abnormalities within helper T cells, regulatory T cells, and monocytes from the peripheral blood of ER+ BC patients that reflect immune activity within tumors and can predict clinical outcome. Furthermore, our early data showed that cytokines interact in unexpected ways. Application of BNs allows us to distill the complexity of immune signaling and its defects, which involve temporal and spatial dynamics composed of multiple cellular and molecular components, into comprehensive and accurate predictive models that can be further studied in an iterative process between experimental and mathematical approaches.

We believe that information contained in the immune signaling networks in patients with ER+ BC is a critical predictor of clinical outcomes. Accordingly, we are pursuing three research directions:

1.Building data-driven models of the immune signaling network in a control group of healthy individuals using heterogeneous biological data inputs. We are generating experimental data from peripheral blood mononuclear cells (PBMCs) stimulated with multiple cytokines alone or in combination and developing BNs / mathematical models to establish signaling patterns and dynamics in the normal immune system. Biological data inputs include next-gen 30-color flow cytometry, single-cell RNA-sequencing, and functional assays.

2.Building data-driven models of the immune signaling network in ER+ BC patients using heterogeneous biological data inputs. To determine how cancer distorts information flow in immune signaling dynamics, we are using experimental data acquired from patient PBMCs at time of diagnosis and subsequent time points to build networks (using the same approach as above) to determine changes in immune signaling over time and after hormonal treatment. After quantifying intra-and inter-network variability of healthy and ER+ BC patients, we can characterize and model the temporal dynamics of information flow resulting from defects in immune signaling networks.

3.Computational analysis of intratumoral immune infiltration and integration with peripheral blood immune signaling and clinical outcome. We are carrying out high-dimensional histology and spatial image analysis of archival and prospective PBMC-paired human ER+ breast tumors. Subsequently, these data are used to train computational models that predict clinical outcome, then validate this signature and correlate PBMC immune signaling defects with features of paired tumors.

In summary, our work aims to establish a computational framework by which to characterize and study immune signaling as a network and provide new knowledge of that network in healthy and ER+ BC subjects. This work will lead to a comprehensive, multivariate predictor of clinical outcome from peripheral blood, and an improved understanding of defects in cytokine signaling as a novel immune dysfunction of information flow in cancer. An additional deliverable is a computational systems biology data analysis toolkit to construct, interrogate, and dissect signaling networks based on the integration of network and dynamic modeling. These tools are, by design, generalizable to the broader immune-oncology and systems biology applications.

Investigators

Investigators