QP 29 - Response Assessment and Risk Stratification in Gynecologic Cancers: Biomarkers, Imaging, and Outcomes
Presenter(s)
A. Ajao1,2, D. T. Bergman3, J. F. Torres-Roca4, J. G. Scott5, and R. Vargas1; 1Cleveland Clinic Foundation, Cleveland, OH, 2Case Western Reserve University, Cleveland, OH, 3Geisel School of Medicine at Dartmouth, Hanover, NH, 4Department of Radiation Oncology, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, 5Department of Radiation Oncology, Taussig Cancer Institute, Cleveland Clinic Foundation, Cleveland, OH
Purpose/Objective(s):
Cervical cancer radiation dosing remains uniform despite heterogeneous clinical outcomes. The Radiosensitivity Index (RSI) and Genomic Adjusted Radiation Dose (GARD) provide a validated biological framework to quantify radiation effect at a populational level. While RSI/GARD are established in other malignancies, their role in cervical cancer is undefined. Given known outcome differences between squamous and non-squamous histologies, we sought to provide independent populational validation of RSI/GARD in cervical cancer.
Materials/Methods:
RSI was calculated using TCGA cervical cancer RNA-seq data (n = 304). In silico GARD was derived assuming a standard definitive radiation regimen of 75 Gy (2.5 Gy/fraction), enabling assessment of biological dose distribution under uniform physical dosing. To minimize surgical and metastatic confounding, a curated subset (n=49; Stage I-III, no prior hysterectomy) was analyzed for histology-based differences. Gene-level effects for the 10 RSI genes were quantified as standardized mean differences between resistant and sensitive groups to assess translational concordance. RSI was calculated for 17 established cervical cancer cell lines using RNAseq and distributions compared.
Results:
Intrinsic radiosensitivity heterogeneity was observed in TCGA tumors (median RSI 0.33, IQR 0.24-0.46; median in silicoGARD 43.65, IQR 31.14-55.56). Under identical modeled radiation dose, this translated into wide biological effect dispersion across patients, consistent with prior pancancer observations. In the stage I-III subset, RSI and GARD differed significantly by histology (Wilcoxon p = 0.047), independently validating the RSI/GARD framework in cervical cancer. Non-squamous tumors exhibited higher RSI (more resistant; median 0.44) and lower GARD (median 32) relative to squamous histology (median RSI 0.26; GARD 52). Cell lines showed higher median RSI compared to patient tumors (0.50 vs 0.29; Wilcoxon p = 0.0049). RSI gene-level analysis revealed cross-system concordance: ABL1 and AR were associated with resistance, while IRF1, PRKCB, SUMO1, and STAT1 were associated with sensitivity.
Conclusion:
Cervical cancer exhibits marked intrinsic radiosensitivity heterogeneity. Histology-driven RSI/GARD differences are consistent with the clinically observed superior radiotherapy outcomes in squamous disease and support predicted histology-driven radiosensitivity differences. Notably, this separation suggests a cervix-specific finding, as prior pancancer analyses (e.g., lung) have not shown meaningful histology-driven differences and provides a population-level validation aligned with known clinical behavior between subtypes. Despite distributional shifts between in vitro and tumor systems, key resistance drivers remain concordant. These findings extend RSI/GARD validation into gynecologic oncology and support further evaluation of biologically guided radiotherapy strategies in cervical cancer.
