Machine learning may hold promise for analyzing cells co-expressing MYC and BCL2 but lacking BCL6 (M+2+6-) in diffuse large B-cell lymphoma (DLBCL), which can be associated with poorer survival patterns, according to a study presented at the 66th ASH Annual Meeting & Exposition held in San Diego, CA.1
Researchers from the Cancer Science Institute of Singapore, National University of Singapore; the Department of Translational Molecular Pathology at The University of Texas MD Anderson Cancer Center, in Houston; and from other international locations created a scalable workflow to generate and analyze spatial point patterns using multiplexed fluorescent immunohistochemistry data from 449 DLBCL patients across 4 cohorts.
“Malignancies exhibit variable cellular distribution patterns and the relationship between these topographic variations, underlying biological processes, and clinical outcomes remain poorly understood,” the researchers wrote in their abstract. “Point process analyses, widely used in ecology, can elucidate the spatial distribution of points in complex systems but have rarely been applied to tumor heterogeneity…Machine-learning approaches can be applied to understand nuances of cellular point patterns and help with correlating with clinicopathological variables.”
The researchers noted that their workflow leverages Python (OpenCV), which processes the kernel size and intensity for images; QuPath Groovy Script, which automates the image import/export process and parameter thresholding for pixel classification; R Programming, which builds spatial point patterns and annotates oncogene co-expression using GeoJSON formats; and Quality Metrics, which minimizes point exclusion while optimizing spatial accuracy.
The researchers identified 2 M+2+6- spatial phenotypes, which they noted as clustered and dispersed.
They employed a random forest model and with 98% accuracy, classified patients into these groups across all cohorts, and noted that patients with a dispersed phenotype experienced significantly shorter overall survival (P<.05 in 4/4 cohorts).
The researchers noted that their findings underscore the clinical importance of spatial distribution analysis in malignant cell subpopulations.
“We anticipate that this machine-learning pipeline can be developed for clinical use, enabling the classification of spatial phenotypes in DLBCL biopsies for patient stratification,” they concluded in their abstract.
Reference
- Sridhar S, Hoppe MM, Jaynes P, et al. Machine learning classification of spatial patterns of malignant cells reveals implications in prognosis and tumor mircorenvironment composition in lymphoma. Presented at: 66th ASH Annual Meeting & Exposition. December 7-10, 2024. San Diego, CA. Abstract 3603.