Background The SET2,3 index combines an accurate measure of transcription related to both estrogen and progesterone receptors (SETER/PR index) with a baseline prognostic index (BPI) derived from c-T Stage, c-N status and molecular subtype by RNA4 (ESR1, PGR, ERBB2, and AURKA). SET2,3 index was translated from a microarray-based signature to a customized hybridization assay that yields highly reproducible results from routine pathology tissue blocks.
The SET2,3 index was calculated from microarray data prepared from baseline biopsies of HR+/HER2- cancers prior to neoadjuvant taxane-anthracycline chemotherapy followed by adjuvant endocrine therapy in: 1) a test set of 307 samples from the MDACC cohort with 8 years median follow-up (Affymetrix U133A microarrays, Symmans et al JAMA 2011), and 2) an independent blinded validation set of 268 high-risk Mammaprint (MP) samples from the I-SPY2 clinical trial with 3.8 years median follow-up (Agilent 44K microarrays, Agendia). The cut-point for high versus low SET2,3 was defined in the MDACC cohort and then tested in the I-SPY2 cohort. Association between the SET2,3 index and distant relapse-free survival (DRFS) was evaluated in each cohort using multivariate Cox regression models. One model adjusted for residual cancer burden (RCB) after neoadjuvant chemotherapy and its interaction with the SET2,3 index, also including treatment arm (experimental vs. control) in the I-SPY 2 trial. Another model adjusted for other prognostic gene expression signatures (GES) for recurrence score (RS), MP, or endopredict (EP) in the MDACC cohort, and actual MP in the I-SPY2 trial. A third model tested the ability of prognostic GES to substitute for RNA4 subtype (part of BPI) or SETER/PR index as components of the SET2,3 index.
Predicted sensitivity to endocrine therapy (SET2,3) and response to neoadjuvant chemotherapy (RCB) were independently prognostic, with nonsignificant (NS) interaction term, in multivariate analyses of the MDACC study: SET2,3 HR 0.26 (p=0.016); RCB HR 2.04 (p<0.001); and validated in the I-SPY2 trial: SET2,3 HR 0.27 (p=0.031); RCB HR 1.68 (p=0.008); treatment arm HR 1.03 (NS). Other GES (RS, MP, EP) were prognostic in univariate analyses, but were NS in multivariate models with SET2,3 index. They could each effectively substitute for RNA4 subtype, but not for SETER/PR index, as a component of SET2,3 index. SET2,3 index (HR 0.39, p=0.002) was prognostic independently of MP score (HR 2.82, p=0.299) in multivariate Cox analysis of the I-SPY2 trial. RCB classes of chemotherapy response were strongly prognostic in cancers with low SET2,3 (MDACC p<0.001, I-SPY2 p<0.001) but were NS in cancers with high SET2,3 (frequency: 41% MDACC, 37% I-SPY2).
SET2,3 index of endocrine transcriptional activity in a pre-treatment core biopsy added independent significant prognostic information to any baseline prognostic score and response to neoadjuvant chemotherapy. Approximately 40% of clinical or genomic (MP) high-risk HR+/HER2- cancers had high SET2,3 index, and their response to neoadjuvant chemotherapy (RCB class) was not prognostic. This suggests that SET2,3 index might be used to select appropriate candidates for endocrine-based neoadjuvant treatments in clinical trials.