Translat Oncol

Real-Time Measurement of Functional Tumor Volume by MRI to Assess Treatment Response in Breast Cancer Neoadjuvant Clinical Trials: Validation of the Aegis SER Software Platform

Newitt DC, Aliu SO, Witcomb N, Sela G, Kornak J, Esserman L, Hylton NM

PURPOSE: To evaluate the Aegis software implementation for real-time calculation of functional tumor volume (FTV) in the neoadjuvant breast cancer treatment trial setting.

METHODS: The validation data set consisted of 689 contrast-enhanced magnetic resonance imaging (MRI) examinations from the multicenter American College of Radiology Imaging Network 6657 study. Subjects had stage III tumors ≥3 cm in diameter and underwent MRI before, during, after receiving anthracycline-cyclophosphamide chemotherapy. Studies were previously analyzed by the University of California San Francisco core laboratory using the three-timepoint signal enhancement ratio (SER) FTV algorithm; FTV measurement was subsequently implemented on the Hologic (formerly Sentinelle Medical Inc) Aegis platform. All cases were processed using predefined volumes of interest with no user interaction. Spearman rank correlation was evaluated for all study sites and visits. Cox proportional hazards analysis was used to compare predictive performance of the platforms for recurrence-free survival (RFS) time.

RESULTS: Overall agreement between platforms was good; ρ varied from 0.96 to 0.98 for different study visits. Site-by-site analysis showed considerable variation, from ρ = 0.54 to near perfect agreement (ρ = 1.000) for several sites. Mean absolute difference between platforms ranged from 1.67 cm3 pretreatment to 0.2 cm3 posttreatment. The two platforms showed essentially identical performance for predicting RFS using pretreatment or posttreatment FTV.

CONCLUSION: Implementation of the SER FTV algorithm on a commercial platform for real-time MRI volume assessments showed very good agreement with the reference core laboratory system, but variations by site and outlier analysis point out sensitivities to implementation-specific differences

doi link