PQA 07 - PQA 07 Gastrointestinal Cancer and Sarcoma/Cutaneous Tumors Poster Q&A
2998 - Modeling Radiosensitivity with Gene Interactions Improves the Prediction of Pathological Response to Neoadjuvant Concurrent Chemoradiation in Esophageal Squamous Cell Carcinoma
F. M. Hsu1, J. C. H. Cheng2, C. H. Hsu3, Y. L. Chang4, and J. M. Lee5; 1Department of Radiation Oncology, National Taiwan University Cancer Center, Taipei, Taiwan, 2Division of Radiation Oncology, Department of Oncology, National Taiwan University Hospital, Taipei, Taiwan, 3National Taiwan University Cancer Center, Taipei, Taiwan, 4Department of Pathology, National Taiwan University Hospital, Taipei, Taiwan, 5Department of Surgery, National Taiwan University Hospital, Taipei, Taiwan
Purpose/Objective(s): Pathological complete response (pCR) following neoadjuvant concurrent chemoradiation (neoCCRT) significantly improves prognosis and increases the likelihood of organ preservation in patients with locally advanced esophageal squamous cell carcinoma (LA-ESCC). The radiosensitivity index (RSI), compromising ten pivotal genes involved in double-strand DNA repair, has been proposed to predict radiotherapy response. However, its application in LA-ESCC, especially when utilizing clinical Formalin-Fixed Paraffin-Embedded (FFPE) samples, has been scarcely explored. We aimed to evaluate the performance of RSI score and models in predicting pCR for LA-ESCC patients receiving neoCCRT, potentially enabling personalized treatment strategy that could lead to improved outcomes and reduced treatment-related morbidity.Materials/
Methods: Patients with resectable LA-ESCC were prospectively enrolled. They received 40–45 Gy of radiotherapy in 20–25 fractions plus concurrent platinum-based chemotherapy, followed by radical esophagectomy. RNA isolated from FFPE samples of endoscopic biopsies was analyzed using a novel reverse transcription polymerase chain reaction-based hybridization protection reaction technique to measure expressions of RSI genes (AR, JUN, STAT1, PRKCB, RELA, ABL, SUMO1, PAK2, HDAC1, and IRF1). pCR was defined as no residual primary tumor or metastatic lymph nodes in esophagectomy specimens. Predictive models for pCR, considering both individual RSI genes and gene-gene interactions, were built using multiple logistic regression and evaluated by the area under the receiver operating characteristic curve (AUROC). The chi-square test, Mann-Whitney test, and Kaplan-Meier method were used for statistical analysis. Results: Sixty patients were enrolled, with 18 (30%) achieving pCR after neoCCRT. No significant difference in median RSI scores was observed between the pCR and non-pCR groups (0.31 vs. 0.52, p=0.15). A cutoff value of 0.5 distinguished patients with high (n=23) and low (n=37) RSI groups, with high RSI patients showing a better pCR rate (43.5% vs. 21%, p=0.07), superior overall survival (median not reached vs. 25.2 months, p=0.02), and progression-free survival (median 72 months vs. 24 months, p=0.03). Among all RSI genes, HDAC1 was most significantly associated with pCR. The predictive model incorporating gene-gene interactions demonstrated a higher AUROC (0.898) compared to the model treating each gene as an independent factor (0.802, p=0.19). Conclusion: The expression of RSI hub genes in clinical FFPE samples is associated with pCR and improved clinical outcomes following neoCCRT for LA-ESCC. Predictive models that incorporate gene-gene interactions outperform those considering each gene independently. These results warrant further validation in a larger, independent cohort, potentially refining patients treatment strategy (e.g., bi-modality vs. tri-modality) based on precision oncology.
