Objective: This prospective observational study aimed to compare ¹⁸F-fluorodeoxyglucose positron emission tomography-based metabolic response with Response Evaluation Criteria In Solid Tumors (RECIST)-based response after two cycles of platinum-based chemotherapy in predicting clinical outcome of patients with advanced non-small cell lung cancer (NSCLC).
Methods: Untreated patients with advanced NSCLC scheduled to receive platinum-based chemotherapy were enrolled for this study. They underwent spiral computed tomography and ¹⁸F-fluorodeoxyglucose positron emission tomography concurrently before and after two cycles of chemotherapy. An optimal maximum standardized uptake value reduction of primary lesion was investigated retrospectively for a metabolic response.
Results: A total of 43 patients were eligible for final analysis from August 2003 to May 2007. Objective response rate (ORR) was significantly higher in RECIST-based responders compared with RECIST-based nonresponders after two cycles of platinum-based chemotherapy [85.0% (17/20) vs. 4.3% (1/23), respectively; P=0.000], but median progression-free survival (PFS) and median overall survival (OS) were similar [5.8 95% confidence interval (CI): 3.6-8.0) vs. 5.0 (95% CI: 3.3-6.7) months, respectively, P=0.761; 13.7 (95% CI: 7.3-20.1) vs. 15.5 (95% CI: 3.8-27.2) months, respectively, P=0.356]. At the optimal cut-off value (maximum standardized uptake value reduction of primary lesion by 31%) for a metabolic response after two cycles of chemotherapy, metabolic responders had a significantly higher ORR [66.7% (16/24) vs. 10.5% (2/19), P=0.000] and a longer median PFS [6.5 (95% CI: 5.2-7.8) vs. 4.8 (95% CI: 2.9-6.7) months, P=0.041]; median OS was 17.7 months (95% CI: 9.2-26.2) in metabolic responders and 12.0 months (95% CI: 3.3-20.7) in metabolic nonresponders (P=0.799).
Conclusion: Metabolic response could be superior to RECIST-based response after two cycles of platinum-based chemotherapy in predicting PFS of untreated patients with advanced NSCLC. Both of them could be predictive for ORR, but neither of them could predict OS.