Correlation between Magee score and recurrence score in Indian patients with breast cancer.

Abstract

e12555 Background: Recurrence score (RS) from Oncotype Dx (ODx) test predict chemotherapy benefit in estrogen receptor (ER) positive, and HER2-negative patients with breast cancer. ODx is an expensive test in an Indian practice setting - cost of adjuvant chemotherapy is one-sixth of the cost of the test. Magee equations use histopathological variables, and semiquantitative results from ER, PR and Ki-67, to calculate Magee score (MS) and provide a cost-effective surrogate mechanism for predicting RS. We aimed to assess the correlation between MS and RS in an Indian setting. Methods: Pathology reports were obtained from oncologists in India. Immunohistochemistry for ER and PR were reported using the Allred system. Therefore, two methods were identified to estimate H-scores - using a calculated method (A), and using an expert-approximated (B) method. MS were calculated by a single investigator blinded to the RS. MS and RS were then assessed for correlation via the Pearson correlation method. MS were arbitrarily stratified into ≤ 15, 15 - 30, and ≥ 30, and their overall agreement (OA) or concordance with RS risk categories, ≤ 15, 15 - 30, and ≥ 30, were assessed. The OA was assessed by determining the proportion of patients in similar risk categories across low, intermediate and high-risk RS and MS. Results: Among 34 patients included in the study, 38% (13) had an RS ≤ 15, 38% (13) had an RS between 15 and 30, and 23% (8) had an RS ≥ 30. Correlation coefficient between MS and RS was 0.78 for both methods of estimation (p<0.0001). The OA between MS-A and RS, and MS-B and RS were 59% (20/34) and 47% (16/34) respectively. The concordance between low-risk MS-A and RS is 78% (n=9) and between low-risk MS-B and RS is 50% (n=20). The concordance between high-risk MS-A and RS is 100% (n=2) and high-risk MS-B and RS is 100% (n=2). Conclusions: Although there is good correlation between RS and MS, the concordance between the two in similar risk categories is not optimal. Of the two methods to estimate H-score from Allred score, the calculated method seems to be more optimal. Our pilot study shows good correlation between MS and RS. We recommend that pathologists incorporate H-score as part of the reporting of ER and PR. A larger prospective study will be useful to test our hypothesis for the use of MS as a predictive tool for adjuvant chemotherapy in breast cancer.