Abstract
BACKGROUND: Oncotype DX (ODX) is widely used to estimate recurrence risk and guide adjuvant therapy in hormone receptor-positive (HR+), HER2-negative early-stage breast cancer. However, limited accessibility and high costs have prompted the use of alternative clinical models, such as the Tennessee nomogram. This study aimed to validate the predictive performance of the Tennessee nomogram in a Korean breast cancer cohort and identify factors contributing to discrepancies between nomogram predictions and ODX results. METHODS: We retrospectively analyzed data on1298 patients with HR+/HER2-, node-negative invasive breast cancer who underwent ODX testing between May 2013 and August 2023. Predictive probabilities were calculated using the Tennessee nomogram and compared with actual ODX recurrence scores. Sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV), and area under the curve (AUC) were determined. Discordant cases were examined for clinicopathologic characteristics contributing to prediction errors. RESULTS: The nomogram demonstrated an overall accuracy of 86.1% (sensitivity 0.130, specificity 0.989, AUC 0.776). Discordant results were observed in 13.9% of cases, primarily in patients with a high histologic grade, PR negativity, and elevated Ki-67 index. Most false negatives clustered within the ODX score range of 25-30, suggesting underestimation of risk in borderline-high cases. CONCLUSIONS: The Tennessee nomogram may be a useful surrogate when ODX testing is unavailable, but caution is warranted in patients with aggressive tumor biology. In such cases, ODX testing should be prioritized to guide adjuvant therapy decisions.