Imaging-based patient inclusion model improves clinical trial performance

Abstract

BackgroundQuantitative image analytics (“radiomics”) is a powerful tool for predicting and prognosing cancer patient outcomes in response to therapy. We hypothesize that radiomic features would be useful as inclusion/exclusion criteria for patient enrichment in clinical trials and aimed to develop the appropriate framework for this analysis.MethodsThis was tested among soft-tissue sarcoma (STS) patients accrued into a randomized clinical trial (SARC021) that evaluated the efficacy of evofosfamide (Evo), a hypoxia activated prodrug, in combination with doxorubicin (Dox). Notably, SARC021 failed to meet its survival objective. We tested whether a radiomic biomarker-driven inclusion/exclusion criterion could have been used to result in a significant treatment benefit of the Evo+Dox combination compared to the standard Dox monotherapy. A total of 166 radiomics features were extracted from 303 patients from the SARC021trial with lung metastases, divided into a training and test set. Univariable and multivariable models were utilized to discriminate OS in the two treatment groups.FindingsA single radiomics feature, Short Run Emphasis, was the most informative. When combined with histological classification and smoking history, an enriched subset (42%) of patients had longer OS in Evo+Dox vs. Dox groups [p=0.01, Hazard Ratio (HR) =0.57 (0.36-0.90)]. Application of the same model and threshold value in an independent test set confirmed the significant survival difference (p=0.002, HR=0.29 (0.13-0.63)). This process also identifies patients most likely to benefit from doxorubicin alone.InterpretationThe study presents a first of its kind radiomic approach for patient enrichment in clinical trials. In particular, we have shown that had the radiomic model been used for selective patient inclusion into the SARC021 trial, it would have met its primary survival objective for patients with metastatic STS.

Predicting Tumour Response With Radiomics and Machine Learning in MR-Guided Cervix Brachytherapy

“…This would necessitate the use of Randomized Control Trials (RCTs). Incorporating radiomic features into the inclusion criteria of RCTs is currently an emerging trend in radiomics research [124,125,126].…”

Predicting Tumour Response With Radiomics and Machine Learning in MR-Guided Cervix Brachytherapy

“…This would necessitate the use of Randomized Control Trials (RCTs). Incorporating radiomic features into the inclusion criteria of RCTs is currently an emerging trend in radiomics research [124,125,126].…”