Thesis or Dissertation A combination of preoperative CT findings and postoperative serum CEA levels improves recurrence prediction for stage I lung adenocarcinoma

Yamazaki, Motohiko  ,  山崎, 元彦

2015-03-23 , 新潟大学
学位の種類: 博士(医学). 報告番号: 甲第3976号. 学位記番号: 新大院博(医)甲第622号. 学位授与年月日: 平成27年3月23日
European Journal of Radiology. 2015, 84(1), 178-184.
Objectives: To assess the prognostic value of combined evaluation of preoperative CT findings and pre/postoperative serum carcinoembryonic antigen (CEA) levels for pathological stage I lung adenocarcinoma. Methods: This retrospective study included 250 consecutive patients who underwent complete resection for <__-3-cm pathological stage I (T1–2aN0M0) adenocarcinomas (132 men, 118 women; mean age, 67.8 years). Radiologists evaluated following CT findings: maximum tumor diameter, percentage of solid component (%solid), air bronchogram, spiculation, adjacency of bullae or interstitial pneumonia (IP) around the tumor, notch, and pleural indent. These CT findings, pre/postoperative CEA levels, age, gender, and Brinkman index were assessed by Cox proportional hazards model to determine the best prognostic model. Prognostic accuracy was examined using the area under the receiver operating characteristic curve (AUC). Results: Median follow-up period was 73.2 months. In multivariate analysis, high %solid, adjacency of bullae or IP around the tumor, and high postoperative CEA levels comprised the best combination for predicting recurrence (P < 0.05). A combination of these three findings had a greater accuracy in predicting 5-year disease-free survival than did %solid alone (AUC = 0.853 versus 0.792; P = 0.023), with a sensitivity of 85.7% and a specificity of 74.3% at the optimal threshold. The best cut-off values of %solid and postoperative CEA levels for predicting high-risk patients were >__-48% and >__-3.7 ng/mL, respectively. Conclusion: Compared to %solid alone, combined evaluation of %solid, adjacency of bullae or IP change around the tumor, and postoperative CEA levels improves recurrence prediction for stage I lung adenocarcinoma.

Number of accesses :  

Other information