Revolutionizing Pathology Practices For Improved Precision Oncology: Leveraging AI to Enhance Efficacy of Immunotherapy

This is part six of our report on the Digital Pathology & AI Congress: USA. For other parts, please see links toward the bottom of the article. The conference featured insightful discussions on the practical aspects and potential impact of machine learning applications in pathology, showcasing the potential of AI to enhance diagnostic accuracy and efficiency, improve patient care, and foster personalized medicine. Parts five through eight will cover these discussions.

 

Tamara Jamaspishvili of SUNY Upstate Medical University discussed the transformative potential of AI and computational pathology in addressing gaps in current pathology practices and enhancing the efficacy of immunotherapy by improving the accuracy and consistency of biomarker assessment.[1]

Dr. Jamaspishvili showed clinical data supporting the idea that Al-based quantitative approaches in biomarker assessment are crucial for advancing companion diagnostic testing and personalized medicine by improving the risk stratification of patients with prostate cancer. In a multicenter retrospective study, Dr. Jamaspishvili’s team showed that AI-based testing for PTEN loss could predict metastasis after surgery in patients with prostate cancer. She argued that AI-based risk stratification of patients could inform personalized treatment approaches, thereby improving patient outcomes.

PD-L1 scoring is required for patient selection for immunotherapy. However, manual scoring is labor intensive, has high inter- and intra-observer variability, and lacks standardization. Dr. Jamaspishvili discussed recent data showing that AI-powered quantification of PD-L1 expression may improve biomarker scoring and identify more patients who would benefit from treatment with immune checkpoint inhibitors.

She also showed unpublished data from her recent work that aims to develop AI models to predict TME features in patients with non-small cell lung cancer. They used attention-based multiple instance learning to analyze H&E-stained whole slide images and identified two predicted clusters based on 30 gene signatures. The first cluster contained patients with immune-inflamed TME, whereas the second cluster included patients with immune-desert TME. They then used supervised machine learning to identify the most important TME features and predict treatment responses. Validation of their models on 231 patients suggests that the use of Al to predict TME features from H&E images is a promising and cost-effective method for assessing biomarker expression with high accuracy, improving the prediction of immune responses and aiding in patient stratification and treatment selection.

However, high costs and reimbursement uncertainties are important barriers to the clinical adoption of digital pathology. Dr. Jamaspishvili concluded that value-based care becomes increasingly evident in the clinical adoption of Al algorithms, and these data provide strong evidence for future reimbursement policies.

Dr. Jamaspishvili, Assistant Professor and Director of the Department of Pathology Research Core & Digital Pathology, SUNY Upstate Medical University.

Links To Other Parts Of The Series

Part 1: Highlights from the 10th Digital Pathology & AI Congress: USA

Part 2: Digital Pathology Implementation: Insights From Experts At DP&AI: USA

Part 3: Clinical Implementation Challenges And Potential Solutions

Part 4: Recent Advances In Digital Pathology

Part 5: Reshaping the Future of Medical Care, Education and Research: The Pivotal Roles of Synthetic Data, Generative AI, and Auto-MLs

References

[1] Tamara Jamaspishvili. Revolutionizing pathology practices for improved precision oncology: Leveraging AI to enhance efficacy of immunotherapy. Presented at the 10th Digital Pathology & AI Congress: USA; May 7-8, 2024; San Diego, CA.

Share This Post

Leave a Reply