AI/ML-Assisted Detection of HMGA2 RNA Isoforms in Prostate Cancer Patient Tissue

Area of Science:

• Computational Biology
• Genomics
• Oncology

Background:

• RNA In Situ Hybridization (RISH) is valuable for spatial gene expression but faces limitations in quantitative analysis due to costly software, especially in under-resourced areas.
• Prostate cancer (PCa) exhibits significant racial disparities, and understanding the molecular drivers, such as High Mobility Group AT Hook-2 (HMGA2), is crucial.

Purpose of the Study:

• To develop an Artificial Intelligence/Machine Learning (AI/ML)-assisted RISH quantification pipeline for analyzing HMGA2 expression patterns in prostate cancer.
• To investigate potential racial disparities in HMGA2 isoform expression within prostate cancer tissues.

Main Methods:

• An AI/ML model was developed to analyze RISH images for quantifying full-length (wild-type) and truncated HMGA2 isoforms.
• Supervised learning analysis was performed on a training dataset generated from RISH images of prostate cancer tissues from 85 men across different racial groups.

Main Results:

• The wild-type HMGA2 isoform was found to be significantly more abundant in tumors from men of African descent compared to other racial groups.
• Wild-type HMGA2 expression positively correlated with increasing Gleason grade, indicating a link to tumor aggressiveness.
• The truncated HMGA2 isoform showed lower abundance and lacked a consistent expression pattern across racial groups.

Conclusions:

• AI/ML-assisted RISH quantification is feasible and can overcome accessibility barriers for spatial gene expression analysis.
• Elevated wild-type HMGA2 expression may serve as a potential biomarker for prostate cancer aggressiveness and contribute to understanding observed racial disparities.
• Equitable computational tools and interdisciplinary collaboration are vital for advancing biomarker discovery and addressing cancer health inequities.