Prediction model for pheochromocytoma/paraganglioma using nCounter assay.

Area of Science:

• Endocrinology
• Oncology
• Genomics

Background:

• The World Health Organization reclassified pheochromocytomas/paragangliomas (PPGL) as malignant in 2017 due to limitations in predicting aggressive behavior.
• Accurate prediction of PPGL metastasis is critical for effective treatment planning.

Purpose of the Study:

• To develop a predictive model for PPGL metastasis.
• To identify key molecular and clinical factors associated with PPGL metastasis.

Main Methods:

• Analysis of 97 PPGL cases from two hospitals, including FFPE tissue samples.
• nCounter assay to identify differentially expressed genes between metastatic and non-metastatic PPGL.
• Development and performance evaluation of a Lasso regression-based prediction model.

Main Results:

• CDK1, TYMS, and TOP2A gene expressions were significantly different between metastatic and non-metastatic PPGL.
• Tumor size showed a positive correlation with CDK1 expression.
• The Lasso regression model achieved 91.7% sensitivity and 95.5% specificity in predicting metastasis.

Conclusions:

• Machine learning-based multi-modal classifiers, utilizing nCounter assay data, can predict PPGL metastasis with high accuracy.
• CDK1 inhibitors represent a potential therapeutic avenue for PPGL treatment.