PET using 18F-FDG is a promising technique to monitor response in oncology. Unfortunately, a multitude of analytic methods is in use. To date, it is not clear whether simplified methods could replace complex quantitative methods in routine clinical practice. The aim of this study was to select those methods that would qualify for further assessment in a future prospective response-monitoring study by comparing results with patient outcome.
Methods: Dynamic 18F-FDG PET scans were obtained on 2 groups of patients. First, 10 patients with advanced non-small cell lung cancer (NSCLC) were scanned on consecutive days before treatment to assess test-retest variability. Second, 30 scans were obtained on 19 patients with locally advanced NSCLC as part of an ongoing response-monitoring study. These scans were analyzed by 2 observers to assess observer variability. In addition, these studies were used to compare various methods with the gold standard, full kinetic analysis (nonlinear regression [NLR]).
Results: Using an image-derived input function, NLR showed excellent test-retest and observer agreement confirming that it could be used as a gold standard method. From a total of 34 analytic methods, 10 showed good correlation with NLR. Taking into account the degree of complexity of the methods, 4 remain for further evaluation.
Conclusion: The optimal method for analysis of 18F-FDG PET data was determined for several levels of complexity. Four methods need to be evaluated further to determine the optimal trade-off between simplicity and accuracy for routine clinical practice.