Concordance of clinical diagnosis of T classification among physicians for locally advanced unresectable thoracic esophageal cancer

Background: We conducted a multicenter phase II trial assessing chemoselection with docetaxel plus 5-fluorouracil and cisplatin induction chemotherapy and subsequent conversion surgery for locally advanced, unresectable esophageal cancer. The aim of this study was to validate the concordance of clinical T diagnosis among physicians in the cases of this trial. Methods: Computed tomography scans and esophagoscopic images of 48 patients taken at baseline were centrally reviewed by 6 senior physicians with experience in esophageal oncology. Individual reviewers voted for definitive T4, relative T4, relative T3, or definitive T3. Discordant diagnoses between reviewers were resolved by the majority opinion. The reviewers were blinded to patient clinical outcome data and to the vote of the other reviewers. Results: Ninety percent of cases were diagnosed as clinical T4 by investigators, while 33.3–75.0% (median 70.8%) of cases were judged to be T4 by 6 reviewers. Discordant diagnosis between investigators and reviewers occurred in 33% (16/48) of all cases (Cohen’s kappa coefficient 0.0519), including 12 cases where curative resection was considered possible (48%, n = 25) and 4 cases where curative resection was considered impossible (17%, n = 23). Critical discordance (one reviewer voted for definitive T3 but the other voted for definitive T4, and vice versa) between reviewers occurred in 0–12.5% of cases (median 2.1%). Conclusions: There were inter-observer variations in clinical diagnosis of the T category of locally advanced, unresectable esophageal cancer. Accurate clinical diagnosis of T classification is required for determining the optimum treatment for each patient.