Novel Radiomic-Based Diagnostic Tool Identifies Patients at High Risk for Prostate Cancer Recurrence

Findings from a new analysis demonstrate “excellent potential” for a developed radiomic-based diagnostic tool to identify patients at high risk for prostate cancer recurrence at time of initial diagnosis.

Prior research has shown multiparametric magnetic resonance imaging (mpMRI)-derived radiomic features to be able to capture subvisual patterns for quantitative characterization of tumor phenotypes in patients with prostate cancer.

Linda M. Huynh, MSc, and colleagues developed, tested, and validated an mpMRI-derived radiomic model for predicting prostate cancer recurrence following initial treatment and presented their work at the 2023 American Society of Clinical Oncology Genitourinary Cancers Symposium. Researchers obtained mpMRI for 76 patients who had undergone radical prostatectomy for treatment of localized disease. All patients had more than 2 years of follow-up with monitoring via prostate-specific antigen, and patients with neo-adjuvant or adjuvant treatment were excluded.

A total of 924 radiomic features were initially extracted, and 75 features (8.1%) were determined to be stable and robust following histogram matching. Additionally, researchers reported that 6 features were determined to be important and nonredundant.

These 6 features were iteratively tested in a training dataset (n=56), and the model with the best parameters yielded a mean receiver-operator curve (ROC) with area under the curve (AUC) of 0.95 ± 0.06.

The model yielded an AUC of 0.67 after application to a validation dataset (n=20). Sensitivity, specificity, positive predictive value, and negative predictive value were 33%, 100%, 40%, and 100%, respectively, researchers added.

“These findings represent excellent potential for the development of a radiomic-based diagnostic tool with high positive predictive value in identifying patients at high risk for prostate cancer recurrence at time of initial diagnosis,” researchers concluded, noting that future projects will incorporate patient demographics and disease characteristics to further improve model sensitivity.

Validation of this model will be pursued with an external cohort of patients.