Abstract
Vast amounts of biomedical data are now routinely available for lung cancer patients ranging from sequencing of lung tissues to liquid biopsies. In addition, new computational tools for quantitatively analyzing immunohistochemistry (IHC) and radiology images are now available. Multi-scale data is now available for complex diseases at molecular, cellular and tissue scale to establish a more comprehensive view of key biological processes. Intra and inter individual heterogeneities are often quoted as the main challenge for studying cancer. These heterogeneities exist at all scales, from microscopic to macroscopic. However, the technologies used to generate data at a particular scale don’t equally assess the heterogeneity. A single biopsy of a lung tumor does not reflect the potential cellular and spatial heterogeneity. Radiology images on the other hand provide a global picture of the heterogeneity of the complete tumor at the tissue scale.
We propose multi-scale modeling to counter heterogeneity and uncover potentially untapped synergies between different data modalities by integrating information across spatial scales. Multi-scale modeling involves linking information from molecules, cells, tissues, and organs all the way to the organism and the population. We propose to use high dimensional molecular data with cellular and tissue scale image data to develop a statistical multi-scale modeling approach of cancer by tapping the previously understudied information in IHC and radiology images. For example, recent reports showed how to integrate transcriptomic data and CT images resulting in intriguing associations between gene expression signatures and CT image phenotypes. Secondly, multiple studies have shown that EGFR mutation status can be predicted from CT images features, opening up avenues for non-invasive therapy monitoring. In addition, computational analysis of IHC and radiology images is able to predict the overall survival of lung cancer patients. Each of these single reports have shown the potential of multi-scale modeling, however a complete multi-scale model of lung cancer is still under development. Overall, multi-scale modeling can have profound contributions toward predicting diagnosis and treatment by revealing synergies and previously unappreciated relationships. Multi-scale modeling also can contribute to a more fundamental understanding of lung cancer development and can reveal novel insights in how data at different scales are linked to each other.
Citation Format: Olivier Gevaert. Multiscale modeling of lung cancer [abstract]. In: Proceedings of the Fifth AACR-IASLC International Joint Conference: Lung Cancer Translational Science from the Bench to the Clinic; Jan 8-11, 2018; San Diego, CA. Philadelphia (PA): AACR; Clin Cancer Res 2018;24(17_Suppl):Abstract nr IA32.