2018 ASCO Annual Meeting!
Session: Breast Cancer—Local/Regional/Adjuvant
Type: Poster Session
Time: Saturday June 2, 8:00 AM to 11:30 AM
Location: Hall A
Predicting Oncotype DX scores using clinicopathologic features: A report from the National Cancer Database.
2018 ASCO Annual Meeting
Poster Board Number:
Poster Session (Board #43)
J Clin Oncol 36, 2018 (suppl; abstr 551)
Author(s): Catherine Pesce, Kristine Kuchta, Beyza Erdogan, Chihsiung Wang, Katharine Yao, Mahmoud El-Tamer; Department of Surgery, NorthShore University Health System, Evanston, IL; NorthShore University Health System, Evanston, IL; NorthShore University HealthSystem, Evanston, IL; Department of Surgery, NorthShore University HealthSystem, Evanston, IL; Memorial Sloan Kettering Cancer Center, New York, NY
Background: The Oncotype DX recurrence score is used to predict the benefits of chemotherapy added to adjuvant hormone therapy in ER positive early-stage breast cancer. While its use has been validated and cost effectiveness has been established, its expense remains a concern in some health care systems and communities. Using clinical and pathologic features and the National Cancer Data Base (NCDB) we aim to predict patients with low or high Oncotype DX scores. Methods: From 2010-2014, 78,663 breast cancer patients with Oncotype DX scores were selected from the NCDB. Seven clinical and pathologic variables including age, ER, PR, histologic subtype, lymphovascular invasion (LVI), grade, and tumor size were used to predict high-risk ( > 30) or low-risk ( < 18) Oncotype DX scores using logistic regression. Data were split into training (70%) and testing (30%) sets for external validation. The predictive accuracy of the regression model was assessed using a Receiver Operator Characteristic (ROC) analysis. Model fit was analyzed by plotting the predicted probabilities against the actual probabilities. Nomograms were created for visualization of the high-risk and low-risk models using bootstrap estimation method of the model coefficients. Results: Estrogen receptor status, progesterone receptor status, and grade were the strongest predictors of both low-risk and high-risk Oncotype DX scores, followed by age, histology, tumor size, and LVI, yielding AUC of 0.70 for the low-risk model and 0.86 for the high-risk model. Conclusions: We have developed a model that can predict high-risk Oncotype DX scores with very good reliability. Such a tool may be useful in health care systems with limited resources. Model Coefficients of Strongest Predictors of High-Risk Oncotype DX Score
|Tumor size||< 10mm||0.000|
1. TAILORx: Phase III trial of chemoendocrine therapy versus endocrine therapy alone in hormone receptor-positive, HER2-negative, node-negative breast cancer and an intermediate prognosis 21-gene recurrence score.