BACKGROUND: Preoperative treatment is a promising strategy for improving long-term outcomes in advanced esophageal cancer. Two tumor response evaluation criteria for preoperative treatment are available: response evaluation criteria in solid tumors (RECIST) and histological criteria. This prospective study aimed to identify which was a better surrogate end point for survival in the preoperative setting. METHODS: We analyzed all eligible patients (n=164) from the preoperative treatment group in a phase III trial comparing preoperative versus postoperative 5-fluorouracil plus cisplatin for clinical stage II or III esophageal cancer. Intercriteria reliability was evaluated with the proportion of agreement and the kappa coefficient. For validity analyses, hazard ratios (HR) of response to nonresponse and differences in response rates between short- and long-term survivors were evaluated. RESULTS: The clinical and histological response rates were 37.8% (62 of 164) and 20.1% (33 of 164), respectively. The proportion of agreement for response to nonresponse between the 2 criteria was 70.3%, and the kappa coefficient was 0.34. The HR for death in patients with histological response (0.22, 95% confidence interval 0.09-0.55, P<0.001) was lower than for those with RECIST response (0.55, 95% confidence interval 0.33-0.91, P=0.018). The difference in response rates between short- and long-term survivors according to histological criteria (27 vs. 7%, P<0.001) was larger than with RECIST (42 vs. 30%, P=0.13). CONCLUSIONS: Intercriteria agreement was relatively low, and histological criteria yielded more valid assessments of response than RECIST. Histological response rate seemed to be the better surrogate end point of survival in the preoperative setting.
BACKGROUND: Preoperative treatment is a promising strategy for improving long-term outcomes in advanced esophageal cancer. Two tumor response evaluation criteria for preoperative treatment are available: response evaluation criteria in solid tumors (RECIST) and histological criteria. This prospective study aimed to identify which was a better surrogate end point for survival in the preoperative setting. METHODS: We analyzed all eligible patients (n=164) from the preoperative treatment group in a phase III trial comparing preoperative versus postoperative 5-fluorouracil plus cisplatin for clinical stage II or III esophageal cancer. Intercriteria reliability was evaluated with the proportion of agreement and the kappa coefficient. For validity analyses, hazard ratios (HR) of response to nonresponse and differences in response rates between short- and long-term survivors were evaluated. RESULTS: The clinical and histological response rates were 37.8% (62 of 164) and 20.1% (33 of 164), respectively. The proportion of agreement for response to nonresponse between the 2 criteria was 70.3%, and the kappa coefficient was 0.34. The HR for death in patients with histological response (0.22, 95% confidence interval 0.09-0.55, P<0.001) was lower than for those with RECIST response (0.55, 95% confidence interval 0.33-0.91, P=0.018). The difference in response rates between short- and long-term survivors according to histological criteria (27 vs. 7%, P<0.001) was larger than with RECIST (42 vs. 30%, P=0.13). CONCLUSIONS: Intercriteria agreement was relatively low, and histological criteria yielded more valid assessments of response than RECIST. Histological response rate seemed to be the better surrogate end point of survival in the preoperative setting.
Authors: Stanley K T Yu; Aneel Bhangu; Diana M Tait; Paris Tekkis; Andrew Wotherspoon; Gina Brown Journal: Cancer Med Date: 2013-12-16 Impact factor: 4.452