Integrated morphologic analysis for the identification and characterization of disease subtypes

Area of Science:

• Computational pathology
• Digital pathology
• Cancer genomics

Background:

• Disease morphologic variations correlate with molecular events and patient outcomes.
• Quantitative morphometric analysis offers insights into disease mechanisms.
• Image analysis techniques can be used for disease subclassification.

Purpose of the Study:

• To develop a methodology for disease subclassification using image analysis.
• To derive morphologic signatures from digitized whole slide images.
• To demonstrate this methodology with glioblastoma and identify prognostically significant subtypes.

Main Methods:

• Applied methodology to 162 glioblastomas from The Cancer Genome Atlas.
• Generated patient-specific tumor morphology signatures from 462 whole slide images (200 million cells).
• Interrogated morphology-driven clusters for associations with outcome, therapy response, molecular classifications, and genetic alterations; performed genome-wide analysis for transcriptional, epigenetic, and copy number variation events.

Main Results:

• Identified three prognostically significant glioblastoma patient clusters (median survival 15.3, 10.7, and 13.0 months).
• Clustering results were validated in a separate dataset.
• Clusters characterized by molecular events in nuclear compartment signaling, including developmental and cell cycle checkpoint pathways.

Conclusions:

• High-throughput morphometrics can effectively subclassify diseases.
• This approach complements genomic analyses.
• Identified morphology-driven glioblastoma subtypes with distinct clinical and molecular features.