5138 - Radiomic Features as a Non-invasive Biomarker to Predict Response to Immunotherapy in Recurrent or Metastatic Urothelial Carcinoma

Background

Reliable biomarkers to predict response to immunotherapy is crucial for patients’ counseling and decision making. This study was aimed to identify the role of CT radiomic features in predicting response to immunotherapy in patients with recurrent or metastatic urothelial carcinoma.

Methods

A total of 62 patients with their 224 lesions who underwent PD-1 and PD-L1 immunotherapy between March 2015 and November 2017 were retrospectively analyzed. The patients were temporally divided into training sets (n = 41; 155 lesions) and independent test set (n = 21; 69 lesions). For radiomics feature extraction, two radiologists independently segmented the region of interest at baseline CT on portal venous phase. A radiomics signature (RAD score) was built by using the least absolute shrinkage and selection operator (LASSO) method. The diagnostic performance of RAD score for prediction of response to immunotherapy was evaluated by C statistics.

Results

The overall response rate of immunotherapy was 36.6% in the training set and 28.8% in the test set. RAD score revealed the C statistics of 0.83 (95% CI, 0.68–0.93) in the training set and a corresponding C statistics of 0.71 (95% CI, 0.48–0.89) in the test set.

Conclusions

This study suggests that radiomic features extracted from metastatic masses at baseline CT are predictive of response to immunotherapy in patients with recurrent or metastatic urothelial carcinoma.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The author.

Funding

Has not received any funding.

Disclosure

The author has declared no conflicts of interest.