Multiplex Imaging of Localized Prostate Tumors Reveals Altered Spatial Organization of AR-Positive Cells in the Microenvironment

Our recent investigation into the spatial cellular landscape of localized prostate tumors provides a comprehensive analysis of over 20 androgen receptor (AR)-positive immune and non-immune stromal cell types. These cellular populations, many of which have little to no existing information on how AR inhibitors impact their behavior, may play critical roles in tumor progression and response to therapy.

Through advanced spatial profiling techniques, we mapped the tumor microenvironment in treatment-naïve, high-risk prostate cancer patients, aiming to improve patient stratification and inform clinical decision-making. By cataloging these ‘cellular neighborhoods,’ we hope to contribute to the development of more precise tools for risk assessment and treatment planning in prostate cancer.

In response to the growing adoption of spatial profiling methods, we are pleased to present a checklist of best practices for spatial data analysis. This checklist includes:

In silico tissue analysis: Enhances sample size and addresses variability by augmenting the analysis with computational modeling.
Leave-one-patient-out analysis: Evaluates the impact of individual patient variability, which is particularly crucial in studies with small cohorts.
Subsampling analysis: Provides robust estimates of variability in spatial associations, ensuring reliability.
Permutation tests: Assesses the stability of findings, and we recommend utilizing several types of permutation tests to validate results.

By implementing these practices, we aim to standardize spatial data reporting in prostate cancer research, driving more accurate and reproducible findings. Our work offers insights into the molecular and cellular heterogeneity in localized intermediate and high-risk prostate cancer, with the goal of integrating this knowledge into therapeutic decision-making processes.

Written by: Sebnem Ece Eksi, Cancer Early Detection Advanced Research (CEDAR), Knight Cancer Institute, Department of Biomedical Engineering, School of Medicine, OHSU, Portland, OR

Read the Abstract

Login