Computational Oncology and Personalized Medicine COPM2024
Building the world of tomorrow, today

Dr. Kościółek obtained his M.Sc. in chemistry from Jagiellonian University in 2010 and a PhD in biological sciences from University College London (UK) in 2016. In the years 2016-2019 he was a post-doctoral research associate in Rob Knight’s group at University of California San Diego (USA) and 2019-2023 he was a group leader in bioinformatics at the Małopolska Centre of Biotechnology, Jagiellonian University in Kraków. Since 2023 he is a university professor at the Silesian University of Technology in Gliwice. Tomasz is also a co-founder of onebiome, LLC. and an adviser to Human Biome Institute s.a. He is working on the development and application of computational methods to better understand the function and dynamics of the human gut microbiome. He has contributed to the development of a suite of tools commonly used in microbiome analyses - QIIME 2 and Qiita - and he is also working on state-of-the-art machine learning and statistical methods to predict protein function (deepFRI) or to model the dysbiosis and dynamics of the microbiome. The goal of his group is to build a multi-level understanding of the microbiome from genes, through structures, to functions and therapies.

Abstract

The gut microbiome is crucial for human health and disease. Longitudinal studies are gaining importance in understanding its dynamics over time, compared to cross-sectional approaches. Investigating the temporal dynamics of the microbiome, including individual bacterial species and clusters, is essential to understanding the impact of the microbiome on human health. This knowledge has implications for targeted therapeutic strategies, such as personalized diets, probiotic therapy, or the design of biotherapeutics. Using a rigorous statistical framework, we analyze the requirements and limits of modeling microbiome time series. Then, we apply machine learning-based methods to predict microbiome composition over time and as a response to fecal microbiota transplants.

Dr. Tanuj Puri obtained his Ph.D. in medical imaging from University College Dublin in Ireland, in collaboration with King's College London (KCL). He pursued postdoctoral research at the University of Oxford, KCL, and Newcastle University, UK. Currently, he is a postdoctoral researcher at the Department of Cancer Sciences, University of Manchester, focusing on modelling post-radiotherapy toxicities by integrating dose and genomics data. With over 15 years of research experience, Dr. Puri's work has centred on dynamic PET imaging methodology, employing mathematical and statistical models for quantification, image processing, and analysis. His clinical trials have aimed to elucidate the pharmacokinetics of radiotracers such as [18F]FDG, [18F]NaF, [64Cu]CuATSM, and [18F]FMISO for patient-targeted medicine, linking imaging physics with human disease pathophysiology. His contributions include developing methods to measure Ki in bones using shorter PET scan durations, particularly notable for the hip. He has also developed a method to align [64Cu]CuATSM PET images with histology samples and provided insights into cautionary use of [18F]FMISO PET in assessing hypoxia in rectal cancer. Additionally, he has created software tools to measure changes in pulmonary vascular properties in a human model of acute lung injury using DCE-MRI.

Abstract

Objective: Voxel-based dose-response maps (DRM) derived from dose-surface maps (DSM), aim to identify high-risk sub-regions in organs-at-risk (OAR) for studying dose-toxicity relationship. This study assessed the sensitivity of several parameters of DRM pipeline applied to bladder in prostate cancer (PCa) patients undergoing radiotherapy.
Methods: 1808 PCa patients from REQUITE trial were included. The bladders were segmented on planning CT scans, dose surface values extracted, spatially normalised using cylindrical and spherical coordinates with a 91x90 voxel grid, doses converted to equivalent doses in 2 Gy fractions (EQD2) with α/β ratios of 1, 2 and 3 Gy. Five late toxicities (urinary tract obstruction, incontinence, frequency, urgency and retention) were normalised to baseline and dichotomised at grades 1 or 2. Binary analysis of DSMs was performed based on dichotomised toxicity values (0’s & 1’s). DSMs were analysed voxel-wise using Welch's t-test and Mann-Whitney U test with permutation testing with 10, 100, 1000 and 10000 iterations, and the maximum test statistics (Tmax) thresholded at 90th and 95th percentiles to identify statistically significant voxels associated with specific toxicities.
Results: The significant voxels identified on bladder changed with grade dichotomisation (due to changes in the number of events included), coordinate system, statistical test and Tmax thresholding. α/β ratios and permutation testing showed small changes in the results.
Conclusion: The choice of parameter value strongly impacts the identification of high-risk sub-regions on DRMs. All parameters within DRM pipeline should be reported in manuscripts to facilitate external result validation.