Usefulness of computed tomography density of a tumor in predicting the response of advanced esophageal cancer to preoperative chemotherapy

Background In Japan, preoperative chemotherapy is considered essential for resectable stage II or III esophageal cancers. It is important to identify nonresponders for preoperative chemotherapy because continuing ineffective chemotherapy is not beneficial for them. We investigated the correlation between the computed tomography number of tumor and the effect of preoperative chemotherapy in patients with esophageal cancer. Methods This retrospective study included 50 patients receiving preoperative chemotherapy with docetaxel, cisplatin, and 5-fluorouracil for stage II or III esophageal cancer. The computed tomography number of tumor was measured as the mean of Hounsfield Units of the primary lesion on a plain computed tomography measured within a freehand region of interest drawn around the tumor border. For analysis, the patients were classified into responders and nonresponders to chemotherapy, with the pathologic response evaluated using the Japanese and Mandard classification. We analyzed the associations between the computed tomography number of tumor and clinical factors; histopathologic features, including the tumor size, depth of tumor invasion, capillary invasion, Ki-67, p53, and CK5/6 expression; the pathologic response to chemotherapy and prognosis. Results There was a significant association between the computed tomography number of tumor and the response to chemotherapy. The cut-off value of the computed tomography number of tumor in predicting responders to chemotherapy was 40 Hounsfield Units (area under the receiver operating characteristic curve = 0.73, P  = .009); patients with computed tomography number of tumor greater than this value significantly responded to chemotherapy ( P  = .02 in the Japanese and P  = .009 in the Mandard classification) with good postoperative prognosis ( P  = .04). Only Ki-67 expression among the histopathogic features were associated with the computed tomography number of tumor in histopathologic features ( P  = .01). Conclusion The computed tomography number of tumor may be useful to predict the efficacy of preoperative chemotherapy and subsequent prognosis for patients with advanced esophageal cancer.