David van IJzendoorn

LUMC

In short, David van IJzendoorn's fellowship entails the following: 

Every cell counts: understanding leiomyosarcoma subpopulation heterogeneity

'Leiomyosarcoma is a rare cancer with poor five-year survival. There are no accurate predictors for response to conventional chemotherapy modalities, no targeted therapy is available, and immunotherapy has been ineffective. Preliminary data shows that leiomyosarcomas are heterogeneous malignancies, as genomic profiling of multiple regions of primary, recurrent and metastatic tumors from single patients showed extensive intra- and inter-tumor heterogeneity. With the Hanarth Fonds fellowship David van IJzendoorn aims to unravel the extent of the tumor cell heterogeneity on a single cell level, with the goal to identify clinically relevant, subpopulations of cancerous cells and develop new targeted therapies. Leiomyosarcoma samples will be analyzed using advanced single-cell techniques, unsupervised machine learning algorithms will be applied to search for recurrent tumor subpopulations. Subsequently, a machine learning-based analysis will be used to identify targeted therapies for specific subpopulations through drug repurposing. Identified therapies and biomarkers could be rapidly implemented in clinical practice.'

Brief summary of results of this research project February 2023

During my Hanarth Fonds fellowship, I aimed to investigate the tumor heterogeneity of leiomyosarcoma on a single-cell resolution. To achieve this goal, I established a pipeline to process fresh tumor material for single-cell RNA-sequencing (scRNA-seq), machine learning data analysis, and data visualization. With this pipeline, I successfully generated a large dataset of single-cell sequenced samples, including Leiomyosarcoma and other sarcoma subtypes. These data have already resulted in a published manuscript on cell interaction in Tenosynovial Giant Cell Tumors and I presented my work at a leading connective tissue oncology conference. Using scRNA-seq, I identified extensive heterogeneity within leiomyosarcoma samples, including in different metastases from the same patient. To complement the single-cell expression data, I performed bulk RNA sequencing on 96 leiomyosarcoma samples for which extensive follow-up data is available. I developed a deconvolution algorithm trained on the single-cell expression data and used it to perform digital flow cytometry on the bulk RNA-seq samples, and identified a correlation between tumor microenvironment cells, such as macrophages and fibroblasts, and patient disease outcomes. Additionally, I used Random Survival Forests to develop a machine-learning prediction algorithm that correlates the proportions of tumor microenvironment cells with recurrence-free survival.

Brief summary of progress / results January 2022

During the first 11 months of my Hanarth fellowship I have used single cell gene expression data to investigate the different tumor cell populations that are present in leiomyosarcomas (and other tumors). It is likely that specific tumor populations are responsible for metastasis or drug resistance and identifying targeted therapies for those populations will lead to better disease outcome. I have set up a close collaboration with the surgery department to directly process fresh tumor material for single cell sequencing. Up until now I was able to process 10 fresh tumor samples (including leiomyosarcoma, giant cell tumor of bone and tenosynovial giant cell tumors) for single cell sequencing. Using unsupervised machine learning I was able to identify the different cell populations that are present in the samples. This preliminary analysis shows that leiomyosarcomas is a highly heterogeneous disease. To further study the different tumor populations, I have started a collaborating with Enable to perform CODEX on leiomyosarcoma (and other) samples. With CODEX I will be able to study the 2D localization of single tumor cells in the context of their microenvironment.