TIGER Banner
Join

Welcome to TIGER

TIGER is the first challenge on fully automated assessment of tumor-infiltrating lymphocytes (TILs) in H&E breast cancer slides. It is organized by the Diagnostic Image Analysis Group (DIAG) of the Radboud University Medical Center (Radboudumc) in Nijmegen (The Netherlands), in close collaboration with the  International Immuno-Oncology Biomarker working Group (www.tilsinbreastcancer.org).

The goal of this challenge is to evaluate new computer algorithms for the automated assessment of tumor-infiltrating lymphocytes (TILs) in Her2 positive and Triple Negative breast cancer (BC) histopathology slides. In recent years, several studies have shown the predictive and prognostic value of visually scored TILs in BC as well as in other cancer types, making TILs a powerful biomarker that can potentially be used in the clinic. With TIGER, we aim at developing computer algorithms that can automatically generate a "TIL score" with a high prognostic value.

Why breast cancer?

Since 2021, breast cancer is the most common form of cancer worldwide, accounting for 12% of all newly diagnosed cancers. Among women, it is the leading cause of cancer-related death. But not all breast cancers are the same. Depending on the type of cancer, several factors have to be considered when defining the treatment strategy for breast cancer patients, with consequences on the prognosis of patients.

In this context, molecular subtypes are among the most important factors to characterize breast cancer. Four main groups of molecular subtypes are defined in breast cancer literature, based on the status of several receptors used in clinical practice, namely the Hormonal Receptor (HR, which is positive if either Estrogen Receptor (ER) or Progesterone Receptor (PR) are positive) and of their human epidermal growth factor receptor 2 (Her2) :

  • Luminal A (HR positive, Her2 negative)
  • Luminal B (HR positive, Her2 positive)
  • Her2 enriched (HR negative, Her2 positive) 
  • Triple Negative (HR negative, Her2 negative)

Why HER2 and TNBC?

The clinical focus of the TIGER challenge is on Her2 positive and Triple Negative breast cancers (TNBC, negative to all receptors). Studies and clinical evidence show that Her2 positive and Triple Negative breast cancers are the ones with the worst prognosis, and therefore subject of a large corpus of research in prognostic and predictive biomarkers, aiming at improving patient management and prognosis.

Why the TILs?

The treatment of breast cancer in women and men is largely determined by the biology of the tumor. It is becoming more evident that a patient's immunity can be an important indicator of what treatment is needed and how their own immune system can significantly contribute to their chances of long-term survival. In recent years, the role of the tumor microenvironment (TME) has received increasing attention in the immuno-oncology scientific community, with a particular focus on the interaction between tumor cells and the host immune system. In particular, the role of tumor-infiltrating lymphocytes (TILs), immune cells that are part of a person's biology.  TILs are proving to be an important biomarker in cancer patients as they can play a part in killing tumor cells, particularly in some types of breast cancer. Identifying and measuring TILs can help to better target treatments, particularly immunotherapy, and may result in lower levels of other more aggressive treatments, including chemotherapy.

Recommendations from the International TIL Working Group

In 2015, a Working Group was established, with the aim of providing recommendations for the assessment of immuno-oncology biomarkers, including the Tumor-Infiltrating Lymphocytes (TILs). In the context of breast cancer, a seminal paper was published by Salgado et al. [1], proposing a procedure to assess the TILs in H&E-stained breast cancer histopathology slides. Successively, several publications have shown the prognostic (i.e., carrying information about prognosis) and predictive (i.e., carrying information about response to treatment such as chemotherapy) value of TILs visually assessed by pathologists both on surgical resections and on core-needle pre-operative biopsies. One example is the paper of Denkert et al. on Lancet Oncology 2018 [2]. 

The procedure proposed by Salgado et al. is detailed in [1] and supported by educational material available at the https://www.tilsinbreastcancer.org/ web page. The main steps of such a scoring procedure are described in the following video: 


Other approaches to TIL quantification
Next to the recommendations of the TIL Working Group, other approaches could be considered to compute a "TIL score" based on the spatial analysis of immune cells and on morphological patterns detected in H&E whole-slide images. For example, the Immunoscore approach proposed by Galon et al. [3] considers T-cells positive to the CD3 and the CD8 immunohistochemical markers in different locations of tumor region such as the center of the tumor and the border of the tumor. Despite being initially formulated based on IHC and for colorectal cancer, the Immunoscore is an additional example of the importance of the spatial organization of the TILs and a similar approach could be developed to quantify immune infiltrate in H&E slides as well. Other studies have been also presented, related to breast cancer [4], but also to lung cancer [5] and more in general in a pan-cancer approach [6], showing the prognostic value of the TILs computed using several approaches. Additional information on the scientific evidence of the role of the TILs can be found here.

The goal of the TIGER challenge

The goal of this challenge is two-fold. First, allow the scientific community to develop AI models for automated quantification of TILs in TNBC and Her2 positive (i.e, Her2 enriched and Luminal B) breast cancer. In this way, we aim at enabling the creation of open AI-based solutions that can automate TILs assessment. Second, validate the prognostic value of AI-based automated TILs scores using a large independent test dataset that includes cases from both clinical routine and from a phase 3 clinical trial, which is not directly accessible by participants. We envision the results of TIGER being the first step towards the introduction of automated TIL scoring in clinical practice.

The tasks in the TIGER challenge

Participants in the TIGER challenge will have to develop computer algorithms to analyze H&E-stained whole-slide images of breast cancer histopathology, to perform three tasks:

  1. detection of lymphocytes and plasma cells, which are the main types of cells considered as tumor-infiltrating lymphocytes;
  2. segmentation of invasive tumor and tumor-associated stroma, which are the main tissue compartments considered when identifying relevant regions for the TILs;
  3. compute an automated TILs score, one score per slide, based on the output of detection and segmentation.

Anybody can participate in the TIGER challenge. In TIGER, participants are free to develop their AI-based solution to compute a "TIL score", for example, based on the recommendation of the TIL Working Group, as well as on other approaches such as an Immunoscore-like approach, or based on other types of spatial analysis of lymphocytes and morphological compartments of breast cancer tissue. We expect participants to come up with creative solutions involving the analysis of lymphocytes and plasma cells as well as tissue compartments such as tumor and tumor-associated stroma, to design a TIL-related biomarker with strong prognostic value.  

Via the grand-challenge.org (GC) platform, participants will upload algorithms that will be run by GC on two experimental datasets during the challenge. A first set will be used to assess the computer vision performance of submitted algorithms at segmenting tissue and detecting cells and compiling a leaderboard (leaderboard 1). A second set will be used to assess the prognostic value of automated TILs scores and to compile a second leaderboard (leaderboard 2) (see Evaluation section for details).

For the entire duration of TIGER, both leaderboards will be computed and reported based on the experimental test sets. At the end of TIGER, all algorithms will be evaluated on the final test set, which will be used to compile the final official leaderboards. The final results, used to assign prizes, will be considered as the ones derived from running algorithms on the final test set.

References

[1] R. Salgado et al., "The evaluation of tumor-infiltrating lymphocytes (TILs) in breast cancer: recommendations by an International TILs Working Group 2014", Ann Oncol. 2015 Feb;26(2):259-71

[2] C. Denkert et al., "Tumour-infiltrating lymphocytes and prognosis in different subtypes of breast cancer: a pooled analysis of 3771 patients treated with neoadjuvant therapy", The Lancet Oncology, 19, 2018, 40-50. 


[4] M. Amgad et al., "Joint Region and Nucleus Segmentation for Characterization of Tumor Infiltrating Lymphocytes in Breast Cancer", Proc SPIE Int Soc Opt Eng. 2019 Feb; 10956: 109560M.

[5] K. AbdulJabbar et al, "Geospatial immune variability illuminates differential evolution of lung adenocarcinoma", Nature Medicine, 26, 1054–1062 (2020)

[6] J. Saltz et al., "Spatial Organization and Molecular Correlation of Tumor-Infiltrating Lymphocytes Using Deep Learning on Pathology Images", Cell Rep. 2018 Apr 3;23(1):181-193.e7.