Murat Ak, MD from UPMC speaks about Radiomics analysis for predicting pembrolizumab response in patients with advanced rare cancers.
Link to Abstract:
https://jitc.bmj.com/content/9/4/e001752
Synopsis:
Background information:
A radiomics-based model for predicting pembrolizumab response in patients with advanced rare cancers is presented.
Methodologies:
In the report, 57 patients with advanced rare cancers were enrolled in our pembrolizumab phase II clinical trial. Response Evaluation Criteria in Solid Tumors (RECIST) 1.1 and immune-related RECIST were used to assess tumor response (irRECIST). The patients were divided into two groups: 20 with “controlled disease†(stable disease, partial reaction, or full response) and 37 with “progressive disease.†On standard-of-care, pretreatment contrast improved CT scans, we used 3D-slicer to segment target lesions. Every volume of interest yielded 610 features (10 histogram-based features and 600 second-order texture features). The most unequal characteristics were discovered using least absolute shrinkage and selection operator logistic regression. XGBoost was used to construct a classification model for the prediction of tumor response to pembrolizumab using selected features. To evaluate model results, leave-one-out cross-validation was used.
Conclusions:
In patients assessed by RECIST (94.7 percent, 97.3 percent, and 90 percent, respectively; p0.001) and in patients assessed by irRE, XGBoost-based classification successfully differentiated between controlled disease (complete response, partial response, stable disease) and progressive disease with high accuracy, sensitivity, and specificity. Furthermore, the general features of the RECIST and irRECIST groups accurately, sensitivity, precision, and p meaning of 94.7 percent, 97 percent, 90 percent, p0.001 percent and 96 percent, 96 percent, 95 percent, p0.001 percent, respectively, predicted pembrolizumab response.
Final Thoughts:
In patients with advanced rare cancer, our radiomics-based signature detected imaging variations that predicted pembrolizumab response.
Contextualization:
In patients with advanced rare cancer, our radiomics-based signature detected imaging variations that predicted pembrolizumab response.