Preoperative prediction of a pathologic complete response of esophageal squamous cell carcinoma to neoadjuvant chemoradiotherapy.

BACKGROUND: The accurate prediction of a pathologic complete response (ypT0N0M [LYM] 0 ypStage 0) before operation is essential for selecting appropriate strategies for treating esophageal cancer after neoadjuvant chemoradiotherapy.

METHODS: We reviewed 130 consecutive patients with esophageal squamous cell carcinoma who were evaluated preoperatively using upper gastrointestinal endoscopy, computed tomography, and 18 F-fluorodeoxyglucose-positron emission tomography after neoadjuvant chemoradiotherapy and subsequently underwent esophagectomy. Our aim was to determine the diagnostic abilities of computed tomography, 18 F-fluorodeoxyglucose-positron emission tomography, and endoscopy to predict preoperatively a pathologic complete response of the primary site of the locally advanced esophageal squamous cell carcinoma and associated lymph nodes to trimodal neoadjuvant chemoradiotherapy. Associations between clinical complete response (ycT0N0M [LYM] 0 ycStage 0) and pathologic complete response were investigated preoperatively.

RESULTS: Twenty-nine (22.3%) and 43 (33.1%) patients, respectively, achieved clinical complete response and pathologic complete response, which were associated (P=.001). The sensitivity and specificity, as well as the positive and negative predictive values of clinical complete response to define pathologic complete response were 39.5%, 86.2%, 58.6%, and 74.3%, respectively. Univariate and multivariate analyses selected clinical complete response as the sole independent preoperative predictor of pathologic complete response (clinical complete responses versus non-clinical complete responses: odds ratio: 0.26, 95% confidence interval, 0.10-0.65, P=.004). Recurrence-free and overall survival (OS) rates were better in patients with than in those without clinical complete response (5-year recurrence-free and overall survival: 69.0% vs 41.4% and 75.9% vs 45.0%, respectively, both P=.02). Furthermore, clinical complete response was an independent preoperative predictor of recurrence-free survival (clinical complete response versus nonclinical complete response: hazard ratio: 2.20, 95% confidence interval, 1.08-4.45, P=.03).

CONCLUSION: Although pathologic complete response was predictable preoperatively to some extent, the accuracy was somewhat low. Considerable caution should be exercised when selecting the watch-and-wait approach with operation as needed and omitting planned operative intervention even for patients who achieve clinical complete response after neoadjuvant chemoradiotherapy.