Validated limited gene predictor for cervical cancer lymph node metastases.
Joshua D Bloomstein1, Rie von Eyben1, Andy Chan1
1Department of Radiation Oncology, Stanford University School of Medicine, Stanford, CA, USA.
Oncotarget
|June 30, 2020
Summary
A two-gene model of BIRC3 and CD300LG accurately predicts lymph node involvement in cervical cancer. This gene signature could aid in identifying patients with lymph node metastasis, especially in resource-limited settings.
Area of Science:
- Oncology
- Genomics
- Molecular Biology
Background:
- Lymph node (LN) involvement is a critical prognostic factor in cervical cancer.
- Accurate prediction of LN status is essential for effective treatment planning.
Purpose of the Study:
- To identify differentially expressed genes in lymph node-positive (LN+) versus lymph node-negative (LN-) cervical cancer.
- To develop and validate a predictive gene signature for LN involvement in cervical cancer.
Main Methods:
- RNA sequencing was performed on primary tumor biopsies from 74 cervical cancer patients.
- A Random Forest classifier was trained using differentially expressed genes from a training cohort (n=57).
- A 2-gene model (BIRC3 and CD300LG) was developed and validated on a testing cohort (n=17).
Main Results:
- 22 genes exhibited >1.5 fold expression difference between LN+ and LN- groups.
- A 2-gene Random Forest model comprising BIRC3 and CD300LG was identified.
- The model achieved 88.2% classification accuracy and a 98.6% ROC-AUC on the validation cohort.
Conclusions:
- A validated 2-gene model (BIRC3 and CD300LG) effectively predicts lymph node involvement in cervical cancer.
- This model holds potential for developing a reverse transcription-quantitative polymerase chain reaction (RT-qPCR) diagnostic tool.
- The tool could assist in identifying patients with LN involvement, particularly in resource-limited settings.
