Scientific Annual Report 2020



Faster turn-around time in clinical trials with the help of digital pathology


We developed a digital pathology management platform ‘Slide Score’ in collaboration with software engineer Jan Hudecek (Research-IT) at NKI. It allows sharing, annotating, and efficient data collection from digital pathology slides. Slide Score has been integrated into the research workflow within the Core Facility Molecular Pathology and Biobanking. Currently, more than 50.000 whole slides images have been used in about 100 research projects with 400 researchers from 60 institutions. Last year, we applied this software platform as a risk management framework for integration of stromal Tumor-Infiltrating Lymphocytes (sTILs) in a clinical trial (Adaptive phase II randomized trial of nivolumab after induction treatment in triple-negative breast cancer; TONIC trial, led by Marleen Kok) to minimize potential risks of using multi-center pathology assessment to stratify patients in clinical trials (Figure 1).


Using Slide Score for pathology support in the TONIC simplified distribution of slides and collection of assessments from pathologists spread over three different institutions. Using a web browser, pathologists were able to assess the slides at home or on a tablet while traveling which contributed to shorter times needed to get a consensus score needed to stratify patients. Additionally, Slide Score ensured assessments were linked to the slides and the pathologists who made them. All this helped reduce the time patients had to wait before continuing treatment - the median time between biopsy (taking a sample) and assignment to a treatment arm was reduced from the usual 10 days to just over 2 days.


Secondly, we used Slide Score to support research into Artificial Intelligence (AI) and image analysis of the tumor microenvironment. Below you will find two examples:





Scientific Annual Report 2020



Spatial analysis of lymphocytes and fibroblasts identifies biological relevant patterns in estrogen-positive breast cancer


Fibroblasts represent one of the most frequent and plastic cell types in primary breast cancer and both preclinical and clinical research has shown that they harbor intriguing immunomodulatory and architectural capabilities. In estrogen receptor-positive (ER+) breast cancer, higher levels of tumor-infiltrating lymphocytes (TILs) are often associated with a poor prognosis and this phenomenon is still poorly understood. We hypothesized that if the immunomodulatory and architectural capabilities of fibroblasts in a tumor have an impact on TILs (or vice versa), then these features can be used to define spatial tumor clusters with distinct molecular features and prognosis. A deep neural network was developed to locate and identify tumor cells, lymphocytes, and fibroblasts directly from hematoxylin and eosin (H&E) stained whole slides of 935 ER+ primary breast tumors together with a new empirical density estimation analysis to measure the spatial patterns (Figure 2). Using this approach, we found that the spatial analysis of fibroblasts, TILs, and tumor cells on H&E whole slides leads to a new classification within ER+ breast cancer, and these subgroups are associated with differences in survival. Our findings demonstrate a reproducible pipeline for the spatial profiling of breast cancer from H&E whole slides and demonstrate that integrating TILs with fibroblasts in ER+ breast cancer can potentially aid in better stratification of ER+ patients with relatively high TILs in beneficiary or disadvantageous profiles.



Lorem ipsum dolor sit amet, consectetur adipiscing elit. Sed malesuada faucibus ex nec ultricies. Donec mattis egestas nisi non pretium. Suspendisse nec eros ut erat facilisis maximus. In congue et leo in varius. Vestibulum sit amet felis ornare, commodo orci ut, feugiat lorem.




Scientific Annual Report 2020



Unravel the tumor-microenvironment of metastatic HER2-positive breast cancer


Despite major improvements in the treatment of HER2-positive metastatic breast cancer (MBC), only a small fraction of patients achieves complete remission and remain progression-free for a prolonged time. Detailed information on the cancer-immune interactions in HER2-positive MBC is still lacking. To better understand why overall survival differs so widely and which alternative treatment approaches could be of the key for drug development, we characterized the tumor microenvironment in real-world patients with HER2-positive MBC. Infiltrating immune cells and their spatial relationships were characterized by multiplex immunofluorescence and immunohistochemistry. We also evaluated patient samples for immune signatures and other key pathways using RNA sequencing data. With nine years of median follow-up from the initial diagnosis of MBC, we investigated the association between infiltrating immune cells and outcome.



Lorem ipsum dolor sit amet, consectetur adipiscing elit. Sed malesuada faucibus ex nec ultricies. Donec mattis egestas nisi non pretium. Suspendisse nec eros ut erat facilisis maximus. In congue et leo in varius. Vestibulum sit amet felis ornare, commodo orci ut, feugiat lorem.