Abstract P1-02-06: Prediction of Prosigna® breast cancer intrinsic subtype by immunohistochemical ER, PR and Ki67 expression

Abstract The present study evaluated whether Prosigna® breast cancer assay intrinsic subtype (IST) can be predicted using a specified cutoff for Ki67 proliferation index and immunohistochemical (IHC) biological subtype estimation, complemented by estrogen receptor (ER) and progesterone receptor (PR) expression in a clinical cohort. Our study included 185 cases of early stage ER-positive, HER2-negative breast cancer treated at the Comprehensive Cancer Center of the Medical University of Vienna between 2014 and 2019. Prosigna® breast cancer assay was performed when recommended by the institutional multidisciplinary tumor board to aid adjuvant therapy decision. Pathological diagnostics was carried out according to current WHO and EU breast cancer screening and diagnostics guidelines. Immunohistochemical assessment of ER, PR, human epidermal growth factor receptor 2 (HER2) and proliferation index by Ki67 was performed using Ventana Benchmark Ultra (Tucson, AR, USA) according to recent ASCO/USCAP guidelines. Univariable logistic regression was used to verify that Ki67 positivity is predictive for IST LumA and LumB tumors according to Prosigna® IST. Probability for luminal B IST was estimated by receiver operating characteristics and Youden index was calculated. Next, multivariable logistic regression was applied to test the hypothesis that additional inclusion of ER and PR receptor expression contain additional information regarding IST. Median age of patients was 56 years. 52 patients (28%) had tumors larger than 2 cm (pT2-pT3). 75 patients (41%) had lymph node metastases. 83 (45%), 98 (53%) and 3 (2%) of tumors were diagnosed as luminal A, luminal B and HER2-enriched subtype by Prosigna®, respectively. 19 (10%), 77 (42%) and 89 (48%) tumors could be allocated in the low, intermediate and high risk groups, respectively. Univariable logistic regression confirmed that higher Ki67 positivity significantly, yet insufficiently predicts luminal B tumor subtype (p<0.0001). There was no justification for an optimal Ki67 cutoff and irrespective of its value, no more than 70% of tumors could correctly be assigned to the IST, according to a Youden-Index of maximum 0.42. The hypothesis that including ER and PR expression improves accuracy could not be confirmed (p=0.14). Although higher Ki67 positivity significantly predicted luminal B tumor subtype, the Prosigna® breast cancer assay provided independent information for adjuvant therapy decision in cases where Ki67 IST estimation alone is not sufficient. Irrespective of the chosen cutoff value, a maximum accuracy of 70% was achieved for correctly predicting IST, including ER and PR expression into our model did not improve accuracy. Therefore, IST determination by the Prosigna® breast cancer assay cannot be replaced by immunohistochemical parameters. Citation Format: Ulrike Heber, Kristina Tendl-Schulz, Stefan Heber, Sabine Danzinger, Rupert Bartsch, Christian F. Singer, Florian Fitzal, Ruth Exner, Michael F. Gnant, Leonhard Müllauer, Zsuzsanna Bago-Horvath. Prediction of Prosigna® breast cancer intrinsic subtype by immunohistochemical ER, PR and Ki67 expression [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 P1-02-06.