Ioannis Prassas

Pathology is a foundational pillar in patient management. Hence, obtaining an in-depth understanding of disease mechanisms at the tissue/cellular level can unlock critical new insights on disease management, especially in light of the recent advancements in molecular diagnostic platforms and machine learning.

Throughout my training, my research interest has been consistently centered on translational research, specifically related to the elucidation of pathophysiological mechanisms of disease, the identification of novel therapeutic targets and on biomarker discovery. I specialize in molecular diagnostics, with a primary focus on mass-spectrometry-based proteomics, but I have extensive experience utilizing diverse omics and multiplex diagnostic platforms.

Since my appointment as a translational scientist within the Lab Medicine Program at the University Health Network (UHN), I am leading research projects aimed at translating pathology-related discoveries into clinical practice. Leveraging recent advances in molecular diagnostic platforms (spatial transcriptomics, single-cell proteomics, multiplex IHC) my objective is to establish novel ways to mine deeper into pathology data to unlock fundamental biological insights and propel diagnostic discoveries
with immediate clinical utility. LMP (UHN) contains a huge library of annotated biological data that remain largely undermined.

The advent of machine learning and the ongoing switch to digital pathology has opened ample new opportunities for robust pathology data mining. In my new role, my research methodology covers a broad range of activities, starting from ensuring compliance with regulations and extracting pathology data to formulating hypotheses, building diagnostic algorithms, conducting
experiments, and rigorously validating findings. 

Looking forward, I envisage the expansion of our team (namely, “the Integrated Diagnostics Hub”) into a collaborative research laboratory, where interdisciplinary cohorts comprising biology scientists, pathologists and AI specialists meet and synergistically engage in the pursuit of translational insights to revolutionize diagnostic methodologies.

Research synopsis

Area 1: Integrated Multimodal Diagnostics

Despite significant investments in recent decades, few new markers have made it to clinical use in the last decade. As we are shifting away from the idea of finding one "gold biomarker" that works for most patients, there's a new opportunity emerging on the discovery of truly personalized disease markers tailored each person individually. Multiplex omics technologies are becoming more affordable and require an ever-decreasing sample volume for analysis.

With modern molecular diagnostic platforms, we can now mine deeper into examining disease pathways and markers, enabling the development of more precise diagnostics. My current focus lies on the design of personalized and longitudinal biomarker discovery studies towards the development of high-content personalized biomarker profiles.

To this end, we have initiated proof-of-concept studies, particularly in ovarian and renal cancer, with plans to extend our investigations to other malignancies and nonmalignant diseases (e.g. transplantation rejection). This approach holds promise for tailoring treatments and minimizing adverse effects through precise biomarker-guided interventions.

Moreover, recent technological advancements in the field of machine learning have allowed the fusion of vast amounts of high-dimensional biomedical data from multiple sources, including molecular, histopathology, radiology, and clinical records. Early reports highlight a significant improvement in Diagnostic accuracy when multiple layers of info from each patient are meaningfully combined.

I am very interested in developing effective multimodal fusion approaches, leveraging deep learning methodologies to tailor Diagnostics.

Example studies:

1. Longitudinal monitoring of ovarian and renal cancer patients using multiplex immunoassay and NGS-based platforms in pursuit of personalized monitoring proteomic signatures to assess disease progression risk.
2. Fusion of pathomics, radiomics, and molecular patient data to personalize prognosis in testicular cancer

Area 2: Mechanistic/Therapeutic Studies

There is a lot of pathophysiological information hidden deep in a tissue (biopsy). In terms of more basic investigations, my research focus is on the study of tumor microenvironment, with a specific focus on the role of collagen microarchitecture in cancer progression and metastasis.

Building upon some exciting preliminary findings (unpublished) implicating collagenmodifying enzymes in poor prognosis in patients with renal (and other hypoxic) tumors, I am interested in studying the underlying mechanisms involved in the interplay of tumor and stroma that ultimately supports tumor invasion.

Understanding how collagen affects tumor progression by regulating signaling pathways, cytokines, and immune cells in the tumor microenvironment will provide critical insights on the mechanistic role of abnormal collagen deposition (especially enhanced in hypoxic tumors).

Using FFPE-based spatial transcriptomics, quantitative proteomics and multiplex IHC, we are seeking to obtain a better understanding of the tumor-stroma cross-talk associated with the development of pro- fibrotic structures in microenvironment of the tumor that can influence its metastatic potential and the response to systemic treatments (chemo- or immunotherapies).

Example studies:

1. Investigating the therapeutic roles of P4HA2 collagen-related hydroxylase in the metastatic potential of hypoxic tumors
2. Predicting neoadjuvant chemotherapy benefit using deep learning from stromal histology in patients with triple negative breast cancer

Area 3: Digital Pathology/ AI-Driven Diagnostics

The shift from traditional analog pathology (microscopes) to digital pathology (scanned images) marks a significant development in integrated diagnostics. This transition places tissue biology at the forefront of modern diagnostic investigations. Recognizing the emerging impact of digital pathology and artificial intelligence (AI) in revolutionizing diagnostic practices I want to establish collaborative working groups with internal AI scientists to leverage digital pathology platforms and AI algorithms to enhance the accuracy and efficiency of disease diagnosis and unlock new means to personalize predictions.

Over the next three years, we aim to integrate these cutting-edge technologies into our research endeavors, further advancing our ability to identify novel biomarkers, unravel disease mechanisms, and optimize personalized treatment strategies.

Ongoing studies:

1. The development of novel AI-based algorithms to assess response to neo-adjuvant treatments in oncology.
2. The development of deep learning-based methods to augment pelvic lymph node metastasis detection.

Appointments

• Assistant Professor, Department of Laboratory Medicine and Pathobiology, University of Toronto
• Research Scientist UHN