Objective  To evaluate whether clinical imaging findings of sarcomas after preoperative chemotherapy correlate with tumor responses by pathological evaluation using the rate of necrosis, so as to develop reliable and quantitative evaluation of clinical response.

  Methods  We retrospectively reviewed the medical records of 190 patients with high-grade sarcomas (mainly osteosarcomas and Ewing's sarcomas) that originated from the bone and who received neoadjuvant chemotherapy from June 1, 2014 to March 1, 2017 at Peking University People's Hospital. Finally, 157 lesions were evaluated by clinical imaging, including X-ray, computed tomography, magnetic resonance imaging, and bone scans or PET/CT. All patients underwent surgery at our center and pathological evaluation by tumor necrosis rates, which were graded by Huvos' classification, where grade Ⅰ is 0 to 49%, gradeⅡ is 50% to 89%, grade Ⅲ is 90% to 99%, and grade Ⅳ is 100% necrosis. Statistical diversity analysis was performed by different pathological groups and receiver operating characteristic (ROC) curves. ROC curves were generated to determine the dividing clinical parameters (cut-off values) to distinguish different pathological groups.

  Results  The cut-off values of the rate change in maximum diameters of tumors located in the extremities were 86%, 50.7%, and 0.02% for Huvos' Ⅳ, Ⅲ, Ⅱ, and Ⅰgroups, respectively. The differentiation was not obvious using bone scans to distinguish different pathological responses. The cut-off value for SUVmax for Huvos' Ⅲ, Ⅱ, andⅠ groups were 60.7% and 31.4%, respectively. We did not identify any valuable clinical parameters to evaluate the lesion restricted inside the bone. For sarcomas that originated from the axial skeleton, because of the small size of the sample, the differentiation was not so obvious.

  Conclusions  This study clearly defined the measuring methods for sarcomas primarily originating from the bone and attempted to determine meaningful cut-off values for multiple pathological response groups. A prospective multicenter trial is warranted to expand the sample size to make this clinical evaluation more precise and practical.