Prospective validation of the IRSN-23 model for personalized chemotherapy and proposal of new breast cancer subtype classification.

10 Background: The IRSN-23 model uses DNA microarray analysis of tumor tissue to stratify patients into highly sensitive (Gp-R) or less sensitive (Gp-NR) chemotherapy groups based on immune-related gene expression. In a previous report (n=901), Gp-R showed a 39% pCR rate, while Gp-NR had 11% (P=4.98E-23; Sota, Y., et al., Annals of Oncology 25.1 (2014): 100-106). This study aimed to evaluate the reproducibility of the IRSN-23 model in prospective independent validation cohorts and investigate its clinical significance and impact on breast cancer subtype classification. Methods: Tumor tissues were obtained from 143 breast cancer patients undergoing preoperative chemotherapy at Osaka University Hospital (OUH). Patients were classified into Gp-R or Gp-NR using the IRSN-23 model, and its ability to predict pathological complete response (pCR) was assessed. Results were validated with independent public datasets (n=965; breast cancer (n=919) and rectal cancer (n=46)). Results: In the OUH prospective cohort, the pCR rate was significantly higher in the Gp-R group (29.3% (11/41)) than in the Gp-NR group (1.4% (1/71)) without trastuzumab (P=1.70E-5). In all prospective validation cohorts without anti-HER2 therapy, the pCR rate in the Gp-R group was significantly higher than in the Gp-NR group. Pooled analysis of the prospective validation set revealed higher pCR rates in the Gp-R group than in the Gp-NR group, both without (Gp-R 41.9% (104/248) vs. Gp-NR 13.6% (50/367), P=4.0E-15) and with (Gp-R 48.3% (57/118) vs. Gp-NR 35.5% (66/186), P=0.03) anti-HER2 therapy (Table). Collaborative analyses of IRSN-23 Immune Score (IS) and OncotypeDx Recurrence Score (RS) suggested the identification of a highly sensitive chemotherapy group (pCR%; IS>-25+RS>25: 21.6% vs. IS ≤ 25 + RS > 25: 2.9%, P=1.13E-04). PAM50 as an attack factor and IRSN-23 as a defense factor capture the tumor microenvironment, suggesting a new subtype classification of breast cancer. Conclusions: This study offers prospective validation of the IRSN-23 model in predicting chemotherapy efficacy, exhibiting high reproducibility. The findings indicate the IRSN-23 model is clinically valuable and propose new breast cancer subtype classification. These results hold significant implications for personalized treatment strategies.[Table: see text]