Genetic Test Predicts Radiation Toxicity in Prostate Cancer

Genetic Biomarker PROSTOX Validated to Predict Radiation Toxicity in Prostate Cancer Treatment 

A research team at UCLA has validated a microRNA-based germline biomarker that may enhance the personalization of radiation therapy for prostate cancer patients. The biomarker, known as PROSTOX, demonstrated reliable performance in identifying individuals at increased risk for long-term genitourinary (GU) toxicity following stereotactic body radiation therapy (SBRT).

Published in Clinical Cancer Research, the study evaluated 148 participants in the Phase III MIRAGE trial, comparing MRI- and CT-guided prostate SBRT. The results confirmed that PROSTOX—based on inherited microRNA-related single nucleotide polymorphisms (mirSNPs)—can predict late GU toxicity with an area under the curve (AUC) of 0.76. Patients categorized as high risk were found to have an approximately 12-fold increased likelihood of developing significant late GU side effects.

Beyond validating PROSTOX, the researchers identified and characterized three distinct types of radiation-associated GU toxicity: acute only, chronic, and late, each with unique temporal patterns and underlying genetic profiles. These findings were supported through cluster analysis of patient symptom trajectories and genetic modeling using machine learning techniques.

“The ability to differentiate between types of toxicity opens the door for more informed discussions and treatment decisions,” said corresponding author Dr. Joanne Weidhaas. “It brings us closer to tailoring radiation therapy to the individual patient.”

Historically, radiation-related GU side effects in prostate cancer have been grouped as either acute or late, with chronic toxicity—persistent symptoms beyond six months—often overlooked. This study is among the first to demonstrate that chronic toxicity represents a separate entity, both temporally and genetically. Importantly, PROSTOX was not predictive of acute or chronic toxicity, reinforcing the idea that these outcomes stem from different biological mechanisms.

Additional predictive models incorporating mirSNPs, clinical, and dosimetric data improved the ability to identify patients at risk for acute or chronic toxicity, with AUCs of 0.770 and 0.763, respectively. In contrast, models based only on clinical and dosimetric factors performed less well, underscoring the contribution of inherited genetic differences.

Gene ontology analysis further differentiated the biological pathways associated with each toxicity type. Acute toxicity was linked to DNA repair and cellular transport mechanisms, chronic toxicity to cell cycle and apoptosis regulation, and late toxicity—captured by the PROSTOX signature—to immune system activity and inflammation-related signaling.

The study also confirmed that PROSTOX’s predictive performance was not enhanced by adding clinical or dosimetric variables. Notably, it showed no significant correlation with baseline urinary symptoms or other pretreatment factors, suggesting that it captures a distinct genetic susceptibility to late toxicity.

These results suggest that genetic biomarkers like PROSTOX could be integrated into radiation planning workflows to help guide treatment selection. For example, patients identified as high risk for late toxicity could be counseled about alternative treatment modalities with potentially lower risk. Meanwhile, those at risk for chronic toxicity might benefit more from precision techniques like MRI-guided delivery or the use of rectal spacers.

While the sample size of the study was relatively modest, and the concept of chronic toxicity is not yet formally integrated into clinical trial reporting, the findings highlight the value of distinguishing between these toxicity profiles. They also support continued research into genetic predictors of radiation response, with the potential to extend this approach to other cancer types and treatment settings.

“Our findings mark a step forward in identifying patients who are genetically radiosensitive,” said co-author Dr. Amar Kishan. “Ultimately, this could help ensure each patient receives the safest, most effective treatment available.”

Source

Kishan, Amar U., Kristen McGreevy, Luca Valle, Michael Steinberg, Beth Neilsen, Maria Casado, Minsong Cao, Donatello Telesca, and Joanne B. Weidhaas. "Validation and Derivation of MicroRNA-based Germline Signatures Predicting Radiation Toxicity in Prostate Cancer." Clinical Cancer Research, published April 10, 2025. https://doi.org/10.1158/1078-0432.CCR-24-3951.