Concordance between 21-gene score (Oncotype DX) and clinical-pathologic prognostic models in early-breast cancer in Medellín, Colombia.

Abstract

e12521 Background: The genomic-based 21-gene recurrence-score assay (Oncotype DX, Genomic Health)(ODx) is used to decide on the use of adjuvant chemotherapy (ACT) in luminal-type Early-Breast Cancer (LT-EBC). Patients with low RS can safely avoid ACT. Other predictive models based on standard clinical and histopathological (C&H) variables also have been developed. These include, Magee equations (ME), Predict model (PM), and Tennessee nomogram score (TNS). This study aims to establish the concordance between ODx and ME, PM and TNS models in a set of patients with EBC in Medellín, Colombia. Methods: Patients with unifocal, stage I and IIA, LT-EBC (HR+/Her2-) with results for ODx were included. For inclusion, key standard C&H variables needed to be available as to allow accurate assessment of ME, PM and TNS predictive models. ODx was used as the reference test and the predictive models as index tests. Low-risk (LR) was defined < 18 in all three ME; < 3% in the PM; and a probability calculated for LR > 90% in the TNS. A second analysis was performed in the > 50 years-old cohort (+50C), using standard and a modified criteria. In the last one, intermediate-risk (IR) patients were either excluded or grouped with high-risk scores for analysis. Concordance between the models and the ODx was evaluated using Cohen's kappa index (K). The degree of concordance was classified according to the categories established by Landis and Koch. AUROC (area under receiver operating characteristics) was estimated. Statistical analyses were performed with Stata v16. Results: 122 patients were included. Median age: 58 (IQR 49-66). Main histology was ductal carcinoma (85.2%), and median size was 15 mm (IQR 10-20). LR was adjudicated in 80.3%, 57.4%, 89.3% and 69.7% with ODx, ME, PM and TNS, respectively. Concordance between the ODx and ME and PM in the all patient population was fair, with K of 0.35 (95% CI: 0.18-0.50; p < 0.001), and 0.24 (95% CI: 0, 04 - 0.45; p < 0.001), respectively. Concordance of ODx and TNS was inferior with a K of 0.16 (95% CI: 0.03-0.34, p = 0.04). AUC for ME, PM, and TNS was 0.61 (95% CI: 0.49-0.73), 0.61 (95% CI: 0.52-0.70) and 0.59 (95% CI: 0.48-0.70), respectively. 85 patients were included in the +50C (69.7%). For one analysis, 41 patients with IR with either ME or PM were excluded. Concordance between the ME and the ODx was fair. LR in both ME and ODx was found 41/43 patients (95.5%, 95% CI 87.9-100). As for PM and TNS, the degree of concordance with ODx were low, and non-significant, respectively. Concordance between ODx LR was found in 36/37 patients with LR in all three ME, PM and TNS (97.3%, 95% CI: 90.7-100). Discrimination capacity: 79%. Conclusions: The high cost of ODx can be safely avoided in stage I/IIA luminal-type EBC and > 50 year-old with low-risk scores in all three ME, PM, and TNS.