Estimating the OncotypeDX score: validation of an inexpensive estimation tool

Area of Science:

• Oncology
• Genomics
• Biostatistics

Background:

• OncotypeDX is a prognostic and predictive gene expression assay for breast cancer.
• Its clinical utility is limited in resource-constrained healthcare systems.
• There is a need for accessible tools to predict OncotypeDX scores.

Purpose of the Study:

• To develop and validate a simplified model using clinicopathologic criteria to predict OncotypeDX scores.
• To identify patients with low risk of high-risk disease.
• To provide an alternative to expensive genomic assays in limited-resource settings.

Main Methods:

• Retrospective identification of patients with ER/PR-positive, HER2-negative invasive ductal carcinoma who underwent OncotypeDX testing.
• Extraction of clinicopathologic data: tumor size, nuclear and histologic grade, lymphovascular invasion, ER/PR status.
• Development of a simplified risk score using linear regression on a training dataset and validation on a separate dataset.

Main Results:

• Estrogen/progesterone receptors, tumor size, nuclear/histologic grades, and lymphovascular invasion were independently associated with OncotypeDX scores.
• The simplified model correctly assigned 57% of patients to their risk category (<18, 18-30, >30).
• 41% of patients were predicted to have a score <18, with only 2% having true scores >30.

Conclusions:

• A simplified clinicopathologic model effectively predicts OncotypeDX scores.
• This tool can identify a significant proportion of patients with a very low likelihood of high-risk disease.
• The model offers a cost-effective alternative for risk stratification in resource-limited settings.