MRI radiomics captures early treatment response in patient-derived organoid endometrial cancer mouse models

Area of Science:

• Biomedical Imaging
• Oncology
• Radiomics

Background:

• Radiomics extracts quantitative imaging features beyond human visual perception.
• Endometrial cancer research benefits from advanced imaging techniques for treatment response prediction.

Purpose of the Study:

• To develop and validate a radiomic signature (RS) for predicting chemotherapy response in a preclinical endometrial cancer model.
• To assess the efficacy of the radiomic signature in both orthotopic and subcutaneous tumor models.

Main Methods:

• Utilized a patient-derived endometrial cancer organoid model (O-PDX) in mice, treated with chemotherapy or saline.
• Acquired weekly T2-weighted MRI scans and extracted radiomic features using whole-volume segmentation.
• Developed a radiomic model (RS_O) using LASSO statistics and validated it externally (RS_S) on a subcutaneous model.

Main Results:

• The radiomic signature (RS_O) showed high accuracy in predicting treatment response, with AUCs increasing from 0.38 at baseline to 1.0 at endpoint in the orthotopic model.
• External validation (RS_S) demonstrated high accuracy (AUC=0.85) at endpoint in the subcutaneous model, outperforming volumetric MRI (AUC=0.78).
• Neither radiomic signature nor volumetric MRI accurately predicted response at an early timepoint (day 3).

Conclusions:

• A validated radiomic signature effectively captures chemotherapy treatment response in preclinical endometrial cancer models.
• Radiomic tumor profiling via preclinical imaging shows promise for early treatment response assessment in endometrial cancer.
• This approach supports the development of novel therapeutic strategies for endometrial cancer.