Semi-Quantitative Immunohistochemical Scoring System Correlates with Oncotype DX® qRT-PCR Assay for Estrogen and Progesterone Receptors
Wiam Bshara, Li Yan, Song Liu, Thaer Khoury. Roswell Park Cancer Institute, Buffalo, NY
Background: Therapeutic decision-making for women diagnosed with breast cancer requires accurate determination of the estrogen receptor (ER) and progesterone receptor (PR). The cutoff for positive ER/PR is somewhat arbitrary. The aim of the study is to determine the accuracy of Oncotype DX® qRT-PCR assay comparing with IHC and to define IHC cutoff that better predict Oncotype outcome.
Design: Consecutive breast carcinoma cases (n=114) that had Oncotype DX® scoring for ER and PR from 2009 to 2011 were included in the study. IHC for ER and PR (DAKO) was performed on the same tissue block that Oncotype was performed on. Then, ER and PR stains were scored using image analysis (Aperio, Vista, CA) by tagging the tumor cells. The percentage and the intensity stain (0+ to 3+) were recorded. Then, three scoring systems, H-score, Q-score, and Allred-score were calculated. Spearman's rank correlation coefficient test and area under the curve were used for statistical analysis.
Results: ER and PR tested by qRT-PCR were positive in 109 of 114 (97.3%) samples and in 85 of 114 (74.6%) samples, respectively. For ER, there was complete concordance between IHC and qRT-PCR when a cutoff ≥1% was used. One case became discordant when Allred scoring system was used. For PR, 12 samples had discordant results (11 false positive and 1 false negative). Only H-score, similar to qRT-PCR, produced a normal distribution, while Allred and Q-scoring systems produced skewed distribution to the right for both markers. A cutoff for H-score >1.32 for ER and >13 for PR has an accuracy of 100% and 93.9% when compared with qRT-PCR assay, respectively. Although more cases were discordant for PR than ER, the Spearman's rank correlation coefficient was higher for PR than ER (0.88 and 0.63, respectively).
Conclusions: While there is no difference in ER testing by IHC and qRT-PCR, PR has considerable discordance. When referring a breast cancer sample for an Oncotype test, an H-score >1.32 or a percentage ≥1% for ER and H-score for PR>13 can be used as a guide.
Wednesday, March 6, 2013 9:30 AM
Poster Session V # 33, Wednesday Morning