Anant Madabhushi

PhD

Emory University

Robert W. Woodruff Professor, Biomedical Engineering

PLCO (Learn more about this study)

PLCOI-1283

Aug 8, 2023

Improving Risk Stratification in Stage III Colorectal Cancer: An Integrative Analysis of Tumor-Infiltrating Lymphocytes and Tumor Nuclei Features

Our project seeks to build upon previous work conducted in our laboratory focused on the prognostic potential of spatial and density features of tumor-infiltrating lymphocytes (TILs), and tumor nuclei-related features in Stage III colorectal cancer (CRC) patients. Our previous studies have suggested the prognostic value of these features in stratifying patients into distinct risk groups and have shown improved model performance when combining TIL features with tumor nuclei-related features. We aim to refine and validate these models using a larger dataset. Our discovery set only consisted of 50 patients and the current validation set only has 70 patients. Ideally, this expanded dataset would comprise additional Stage III CRC patients with comprehensive clinical data, including progression-free survival, microsatellite status, treatment details, disease free survival and overall survival information. This larger, more diverse patient cohort would allow us to robustly validate our previous findings and refine our models to enhance their prognostic accuracy and clinical utility. In addition, we plan to further investigate the interactions between TILs and tumor nuclei, as our preliminary work suggests these interaction features may carry prognostic value in colon rectal cancer. These features could also be extended to adenocarcinoma from gastrointestinal (GI) tract. By refining the extraction and application of these features, we hope to improve the clinical relevance of our risk stratification models. Ultimately, this work aims to improve the management and outcomes of patients with Stage III CRC and potentially other GI adenocarcinoma.

Our first aim is to expand upon our prior work investigating the prognostic potential of spatial architecture (SpaTIL) of tumor infiltrating lymphocytes (TILs) in Stage III colorectal cancer (CRC) patients. Our previous studies have successfully demonstrated the use of top-5 TIL features in constructing a risk-stratification model with promising results in a validation set of TCGA-COAD patients. The aim is to enhance the robustness and validity of these findings through analysis of a larger, more comprehensive dataset, ideally with comprehensive clinical and pathological data, including progression-free survival, overall survival, microsatellite status, and treatment details. To validate and improve the prognostic performance of our integrative model incorporating both SpaTIL features and tumor nuclei-related features. While our work has shown that combining these two types of features enhanced model performance compared to using SpaTIL features alone, it is critical to further test and refine this approach using a larger, more diverse cohort of Stage III CRC patients.

Our second aim is to explore the interactions between TILs and tumor nuclei, leveraging the analysis of cellular graphs encompassing both cell types. This aim builds on our previous finding that these interaction features carry prognostic value. We aim to refine the extraction and application of these interaction features to enhance their clinical relevance and utility in risk stratification. To enhance the utility of our findings by also including overall survival information in our dataset, where possible. This would allow us to establish more comprehensive

and clinically relevant risk stratification models, potentially improving the management and outcomes of Stage III CRC patients and patients with gastrointestinal adenocarcinoma.

Chuheng Chen (a), Joseph Willis (b), Anant Madabhushi (c),(d),(e)

(a) Department of Biomedical Engineering, Case Western Reserve University, 309 Wickenden Building, 10900 Euclid Avenue, Cleveland OH 44106, USA

(b) Department of Pathology, University Hospitals Cleveland Medical Center and Case Western Reserve University, 11100 Euclid Avenue, Cleveland, OH 44106, USA

(c) Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology and Emory University, 1750 Haygood Drive, HSRBII, N602, Atlanta, GA 30322, USA

(d) Radiology and Imaging Sciences, Biomedical Informatics (BMI) and Pathology, Georgia Institute of Technology and Emory University, 1750 Haygood Drive, HSRB II, N602, Atlanta, GA 30322, USA

(e) Atlanta Veterans Administration Medical Center, 1670 Clairmont Road, Decatur, GA 30033, USA