Repeatability of quantitative whole-body 18F-FDG PET/CT uptake measures as function of uptake interval and lesion selection in non-small cell lung cancer patients

Abstract

Change in 18F-FDG uptake may predict response to anticancer treatment. The PERCIST suggest a threshold of 30% change in SUV to define partial response and progressive disease. Evidence underlying these thresholds consists of mixed stand-Alone PET and PET/CT data with variable uptake intervals and no consensus on the number of lesions to be assessed. Additionally, there is increasing interest in alternative 18F-FDG uptake measures such as metabolically active tumor volume and total lesion glycolysis (TLG). The aim of this study was to comprehensively investigate the repeatability of various quantitative whole-body 18F-FDG metrics in non-small cell lung cancer (NSCLC) patients as a function of tracer uptake interval and lesion selection strategies. Methods: Eleven NSCLC patients, with at least 1 intrathoracic lesion 3 cm or greater, underwent double baseline whole-body 18F-FDG PET/CT scans at 60 and 90 min after injection within 3 d. All 18F-FDG-Avid tumors were delineated with an 50% threshold of SUVpeak adapted for local background. SUVmax, SUVmean, SUVpeak, TLG, metabolically active tumor volume, and tumor-To-blood and -liver ratios were evaluated, as well as the influence of lesion selection and 2 methods for correction of uptake time differences. Results: The best repeatability was found using the SUV metrics of the averaged PERCIST target lesions (repeatability coefficients , 10%). The correlation between test and retest scans was strong for all uptake measures at either uptake interval (intraclass correlation coefficient. 0.97 and R2 . 0.98). There were no significant differences in repeatability between data obtained 60 and 90 min after injection. When only PERCIST-defined target lesions were included (n 5 34), repeatability improved for all uptake values. Normalization to liver or blood uptake or glucose correction did not improve repeatability. However, after correction for uptake time the correlation of SUV measures and TLG between the 60-And 90-min data significantly improved without affecting test-retest performance. Conclusion: This study suggests that a 15% change of SUVmean/SUVpeak} at 60 min after injection can be used to assess response in advanced NSCLC patients if up to 5 PERCIST target lesions are assessed. Lower thresholds could be used in averaged PERCIST target lesions (,10%).