Related Experiment Videos
Unsupervised analysis uncovers changes in histopathologic diagnosis in supervised genomic studies.
Matti Nykter1, Kelly K Hunt, Raphael E Pollock
1Department of Pathology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas, USA.
Technology in Cancer Research & Treatment
|March 23, 2006
Summary
Gastrointestinal stromal tumors (GIST) are distinct from leiomyosarcomas (LMS). Re-evaluating histopathology corrected misclassifications, enabling clear gene expression differences to be identified between these soft-tissue sarcomas.
Area of Science:
- Oncology
- Genetics
- Pathology
Background:
- Human gastrointestinal stromal tumors (GIST) are a distinct mesenchymal tumor type characterized by c-kit mutations.
- Leiomyosarcomas (LMS) of the gastrointestinal tract or retroperitoneum share historical classification with GISTs but have less frequent c-kit mutations.
Purpose of the Study:
- To comprehensively understand the gene expression differences between GIST and LMS.
- To assess the utility of multidimensional scaling (MDS) in differentiating these tumor types.
Main Methods:
- Microarray analysis of 30 GIST and 30 LMS samples.
- Unsupervised analysis using multidimensional scaling (MDS).
- Re-evaluation of histopathology by an experienced pathologist.
Main Results:
- Initial MDS analysis showed poor separation between GIST and LMS groups.
- Re-review of histopathology revealed misclassified cases, particularly GISTs labeled as LMS.
- Corrected diagnoses and relabeling significantly improved separation in MDS analysis, highlighting distinct gene expression profiles.
Conclusions:
- GIST is a distinct sarcoma type, separable from LMS based on gene expression.
- Re-review of histopathology is critical for molecular studies using archival samples.
- Unsupervised analysis (MDS) serves as a vital quality control step in identifying classification issues.