Fluorodeoxyglucose Positron Emission Tomography for Evaluating Early Response During Neoadjuvant Chemoradiotherapy in Patients with Locally Advanced Esophageal Squamous Cell Carcinoma
C. -C. Lin,C. -H. Hsu,K. -Y. Tzen,J. C. Cheng,J. -M. Lee,H. -P. Wang,P. -M. Huang,F. -M. Hsu,W. -W. Chen,T. -C. Huang,A. -L. Cheng
DOI: https://doi.org/10.1016/s0923-7534(20)32047-0
IF: 45.3
2012-01-01
Journal of Clinical Oncology
Abstract:e14576 Background: To determine if fluorodeoxyglucose positron emission tomography (FDG-PET) could differentiate between responding and non-responding locally advanced esophageal squamous cell carcinoma (ESCC) early in the course of neoadjuvant chemoradiotherapy (CRT). Methods: Patients (pts) with T3 or N1M0-1a (AJCC, 6th Ed) ESCC first received a 21-D cycle of induction chemo (paclitaxel 70 mg/m2 D 1, 8; cisplatin 30 mg/m2 D 1, 8; fluorouracil 2000 and leucovorin 300 mg/m2 24-hr D 1, 8). This was followed by concurrent CRT (paclitaxel 35 mg/m2 1-hr D 1, 4 /wk, cisplatin 15 mg/m2 1-hr D 2, 5/wk, and radiotherapy 2 Gy D 1-5 /wk) (Lin CC, et al. Ann Oncol 18:93,2007). When the accumulated radiation dose reached 40 Gy, the feasibility of esophagectomy was evaluated. In pts for whom esophagectomy was not feasible, concurrent CRT was continued to a dose of ≥60 Gy. Serial FDG-PETs were performed before and 14 days after start of induction chemo. Pathologic response (PR) was defined as no or microscopic residual disease. PET response was measured using the standardized uptake value (SUV). Receiver operating characteristic (ROC) analysis was used to evaluate the ability of SUV in distinguishing between PR and non-PR. Results: Of 65 pts (M:F = 60:5, median age 55, range 39-71) with locally advanced (M0:M1a = 58:7) ESCC enrolled from March 2008 to December 2011, 40 pts underwent esophagectomy and 26 pts were pathologic responders. The median SUV decrease 14 days after the start of therapy was 75.8% for pathologic responders and 30.1% for non-responders (p <0.05). In ROC analysis, the area under the curve was 0.70 (95% CI = 0.55-0.86). Using a 35% SUV decrease cutoff value, PET correctly identified 24 of 26 responders (sensitivity 92%) and 8 of 14 non-responders (specificity 57%). The corresponding positive and negative predictive values were 80% and 80%, respectively. Conclusions: A 35% SUV decrease 14 days after the start of induction chemo is significantly associated with PR, but its accuracy in detecting non-responders is too low to justify the clinical use of FDG-PET for early discontinuation of neoadjuvant CRT in pts with locally advanced ESCC.