Poor concordance between 21-gene expression assay recurrence score and all Magee equation recurrence scores in a consecutive cohort of patients treated for hormone receptor positive early breast cancer
Abdel-Rahman Z, Ali H, Favazza L, and Chitale D. Poor concordance between 21-gene expression assay recurrence score and all Magee equation recurrence scores in a consecutive cohort of patients treated for hormone receptor positive early breast cancer. Breast 2017; 32:S87-S87.
The 21-gene expression assay known as Oncotype Dx has become an important tool in the management of hormone receptor positive breast cancer patients. The assay results in a recurrence score (RS), which is used to determine prognosis and estimate the benefit from adjuvant chemotherapy by classifying RS into 3 different categories: low, intermediate and high risk. The Magee equation is a weighted equation which derives a recurrence score using clinical and histological values. Three equations have been created, each incorporates slightly different values to derive a recurrence score (RSm1, RSm2 and RSm3) which follows a 0–100 scale comparable to the 21-gene expression assay RS. Concordance between the equations and Oncotype Dx risk category is 55.8%, 59.4%, and 54.4% for RSm1, RSm2 and RSm3, respectively and 100%, 98.6%, and 98.7% when the intermediate risk category was eliminated. In practice, decisions are not made based on category alone and individual RS are used to determine the risk level, this results in different decisions for patients with different RS within the same category, therefore in order for RSm1, RSm2 and RSm3 to be clinically interchangeable with RS, they must confidently predict individual RS rather than the category. Aim: To determine the concordance between RSm1, RSm2 and RSm3 and individual RS. Methods: 664 patients were included. For each whole number RS, the confidence interval (CI) limits were derived from the standard Oncotype Dx graph. Concordance was defined as RSm falling within the CI of the individual RS for the same patient. Results: 544 patients had the values required for equation 2, 275 each for equations 1 and 3. The median concordances were 33.33%, 28.57% and 28.57% for equations 1, 2 and 3 respectively. Ranges of concordance were identified after eliminating values with 1 data point and again with 3 or less data points; the ranges for the latter were 18–71%, 10–60% and 18–56% for equations 1, 2 and 3 respectively (Table 1)