Prediction of oncotype DX recurrence score by patho-biologic variables and three surrogate models

Abstract Background: The Oncotype DX Recurrence Score (RS) predicts recurrence and chemotherapy benefit in early stage estrogen receptor positive breast cancer patients. Cost and unavailability are two major disadvantages of the assay. Multiple models have been developed to predict the RS using different pathobiologic parameters. The goal of our study was to predict RS based on histopathologic and biomarker features, and to determine concordance and correlation with RS of three surrogate models that are independent of Ki67 index: Breast Cancer Prognostic Score (BCPS), Magee0, and Magee2. Methods: Early stage, ER-positive breast cancer cases with available RS were reviewed (n = 442). ER, PR and HER2 subscores were abstracted from the Oncotype DX reports. RS categories were stratified by pathologic and biomarker variables. Histopathologic and biomarker data were abstracted from pathology reports, and the surrogate RS was calculated by each model. Correlation and concordance between models and actual RS were calculated. Analyses were performed using both conventional (18, 30) and TAILORx (11, 25) thresholds. Results: Less than 5% of breast cancers with pure or mixed lobular features, low grade tumors, carcinomas with high PR content, or Luminal A tumors had a RS >25 (Table 1). Recurrence scores in node-positive tumors were not significantly higher compared to node-negative cases. Both actual and calculated RS were higher in Luminal B versus Luminal A breast cancers. Subscore analysis revealed 99.5% concordance for ER status, 90% concordance for PR status, and 96% concordance for HER2 status (Table 2). BCPS, Magee0, and Magee2 algorithms demonstrated correlation coefficients with RS of 0.63, 0.61, and 0.62, respectively. BCPS showed the best agreement with RS using conventional cutoffs (73%), whereas the two Magee algorithms showed better concordance with the TAILORx thresholds. Two-step discordances were uncommon, especially with the TAILORx cutoffs and Magee2. When a RS of 25 was used to separate high risk from non-high risk cases, concordance rates of 86-88% were achieved for all three models. Using different tumor blocks for Oncotype and biomarker testing did not adversely affect correlation or concordance with RS. Discussion: High RS was observed only in a small percentage of pure or mixed lobular carcinomas, low grade or Luminal A tumors, and tumors with high PR expression, suggesting that these breast cancers may not require Oncotype testing. All three surrogate models demonstrated comparable correlation and high concordance with the RS when a cutoff of 25 was used, suggesting their utility in cases where the actual RS in unavailable. Our data also indicate that models using different hormone receptor quantitation methods (Allred versus H-score) have similar performance characteristics. Possible sources of discordance between actual and computed RS include methodological differences, variable tumor cellularity, intratumoral heterogeneity, and inflammatory infiltrates. Table 1.Stratification of RS Risk Categories by Pathologic and Biomarker VariablesConventional Risk CategoryLow RiskIntermediate RiskHigh RiskTAILORx Risk CategoryLow RiskIntermediate RiskHigh RiskOncotype DX Recurrence Score<1111-1718-2526-30>30n (%)n (%)n (%)n (%)n (%)Histologic Type (n)Invasive ductal carcinoma (340)94 (28)121 (36)73 (21)20 (6)32 (9)Invasive lobular carcinoma (47)10 (21)27 (57)8 (17)1 (2)1 (2)Mixed ductal/lobular carcinoma (22)3 (14)15 (68)4 (18)0 (0)0 (0)Invasive mucinous carcinoma (10)4 (40)4 (40)0 (0)1 (10)1 (10)Molecular Subtype (n)Luminal A (317)112 (35)143 (45)46 (15)11 (3)5 (1)Luminal B (125)9 (7)33 (26)41 (33)10 (8)32 (26)Combined Tumor Grade (n)Low (143)45 (31)59 (41)32 (22)4 (3)3 (2)Intermediate (239)69 (29)101 (42)41 (17)13 (5)15 (6)High (60)5 (8)17 (28)13 (22)5 (8)20 (33)ER (n)High (Allred Score 7-8) (408)120 (29)164 (40)74 (18)21 (5)29 (7)High (Modified H Score ≥ 200) (320)101 (32)133 (42)54 (17)15 (5)17 (5)Low (Allred Score 3-6) (34)1 (3)12 (35)12 (35)1 (3)8 (24)Low (Modified H Score < 200) (122)18 (15)45 (37)33 (27)7 (6)19 (16)PR (n)High (Allred Score 7-8) (306)106 (35)146 (48)41 (13)9 (3)4 (2)High (Modified H Score ≥ 200) (248)96 (39)111 (45)31 (13)8 (3)2 (1)Low (Allred Score 3-6) (97)12 (12)27 (28)32 (33)7 (7)19 (20)Low (Modified H Score 1-199) (155)23 (15)61 (39)41 (26)9 (6)21 (14)Negative (Allred Score 0-2) (39)1 (3)5 (13)14 (36)5 (13)14 (36)Negative (Modified H Score <1) (39)1 (3)5 (13)14 (36)5 (13)14 (36)HER2 (n)Negative (428)120 (28)170 (40)84 (20)20 (5)34 (8)Positive/Equivocal (14)0 (0)6 (34)3 (21)2 (14)3 (21) Table 2.Concordance Between Clinical ER, PR, and HER2 Status and Oncotype DX SubscoreOncotype DX SubscoresClinical LabNegativeBorderlinePositiveERNegative0N/A0Positive2N/A379PRNegative29N/A4Positive34N/A314HER2Negative36540Equivocal800Positive220 Citation Format: Joseph Geradts, Aisha Kousar, Jan Wong, Nasreen Vohra, Mahvish Muzaffar, Anas Mohamed. Prediction of oncotype DX recurrence score by patho-biologic variables and three surrogate models [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P4-06-09.