A combined budding

A combined budding / T-cell score in pT1 and stage II colorectal cancer(CRC)

Numerous immune cell types are found in CRC and have been linked to improved prognosis in CRC; especially T-cell infiltrates have been extensively examined as a protective biomarker.  This proposal aims to build on existing know-how and utilize geometric deep learning methods to create a combined tumor budding/t-cell scoring system.

The results are expected to have a direct impact on patient management and contribute to more precise treatment strategies in patient subgroups.

Lay Abstract

In colorectal cancer, the most important parameter for prognosis and determining patient management is tumor spread according to the Tumor-Nodal- Metastasis (TNM) classification. However, the TNM system may show considerable differences in tumors within the same stage. Therefore, additional biomarkers for a more personalized risk stratification in certain CRC patient groups are needed.
Tumor budding, namely single tumor cells and small clusters of tumor cells at the invasive tumor front, is such a biomarker. However, tumor buds do not account for other factors, such as the body’s protective response. Especially T-cells, a type of inflammatory cell, have been extensively examined as a protective factor. Our own preliminary data demonstrates that a combined assessment of tumor budding and T-cell infiltrates in CRC (termed the Budding/T-cell Score or BTS) may better reflect how aggressive a tumor is than either marker alone.
In this project, we want to further investigate the BTS using graph-based deep learning. On images of slides used to diagnose CRC, graph-based representations are created based on detected tumor buds and T-cells. We believe that this approach will be beneficial for the combined assessment of tumor buds and T-cells, because it captures not only the raw count of the cells but also their structural arrangement, which, we hypothesize, might play a central role for predicting important information for patient treatment. In a third step, geometric deep learning, which extends deep learning beyond the domain of Euclidean geometry, is applied to the graphs to predict clinical endpoints, such as survival time.