Radiologic Response Assessment in Pediatric Soft Tissue Sarcoma: Computed-Assisted Volume Evaluation

Abstract

To compare 3 methods of dimensional assessment, with particular attention to a new software assisted method of volume calculation, in soft tissue sarcoma, and to investigate the interobserver agreement and the intermethod agreement in chemotherapy response classification and resultant clinical repercussions. We studied 34 pediatric patients with nonmetastatic soft tissue sarcoma who had undergone only diagnostic biopsy. Tumor size was measured both at diagnosis and after induction chemotherapy by 3 observers and using 3 measurement methods: maximum axis (1 diameter), estimated volume (3 diameters), and computed volume (software-assisted volume calculation). We used overall concordance correlation coefficient and Bland-Altman statistical methods to assess interobserver agreement and overall concordance correlation coefficient and the κ Cohen coefficient to assess intermethod agreement. According to overall concordance correlation coefficient, the interobserver agreement was very high for each method, with a slight superiority of the software assisted method; this agreement was not confirmed in Bland-Altman plots for maximum axis and estimated volume methods. According to kappa coefficients, the intermethod agreement in chemotherapy response evaluation was poor. Computed volume was the most accurate method in soft tissue sarcoma tumor size assessment. One- and 3-dimensional methods are not concordant in chemotherapy response classification. In particular, the maximum axis method underestimates chemotherapy response and can lead to switching the chemotherapy regimen erroneously.