In addition to tumor size measurements with CT, there is a need for quantitative measurements of metabolic active volumes, possibly adding to tracer uptake measurements in oncologic response evaluation with PET. The aim of this study was to evaluate the metabolic volume test-retest variability in (18)F-FDG and 3'-deoxy-3'-(18)F-fluorothymidine ((18)F-FLT) PET studies for various commonly used volumes of interest (VOIs) and the dependence of that variability on lesion size and relative radiotracer uptake.
Methods: Twenty non-small cell lung cancer patients were scanned twice with (18)F-FDG (n = 11) or (18)F-FLT (n = 9). VOIs were defined on images reconstructed with normalization- and attenuation-weighted ordered-subset expectation maximization using 4 isocontours (A41%, A50%, and A70% thresholds, adapted for local background, and 50% threshold, uncorrected for background). Statistical analysis comprised intraclass correlation coefficients and Bland-Altman analysis.
Results: In the (18)F-FDG and (18)F-FLT groups, 34 and 20 lesions, respectively, were analyzed. Median volumes at the A50% threshold were 3.31 and 2.19 mL (interquartile range, 1.91-8.90 and 1.52-7.27 mL) for (18)F-FDG and (18)F-FLT, respectively. Intraclass correlation coefficients were greater than 0.9, with the exception of the A70%-based metabolic volumes for (18)F-FLT. For lesions greater than 4.2 mL, repeatability coefficients (RCs = 1.96 × SD) of the percentage difference ranged from 22% to 37% for (18)F-FDG and from 39% to 73% for (18)F-FLT, depending on the VOI method being used. Repeatability was better for larger tumors, but there was no dependence on absolute uptake (standardized uptake value).
Conclusion: Results indicate that changes of greater than 37% for (18)F-FDG and greater than 73% for (18)F-FLT (1.96 × SD) for lesions with A50% metabolic volumes greater than 4.2 mL represent a biologic effect. For smaller lesions (A50% VOI < 4.2 mL), an absolute change of 1.0 and 0.9 mL for (18)F-FDG and (18)F-FLT, respectively, is biologically relevant. Considering the balance between the success rate of automatic tumor delineation and repeatability of metabolic volume, a 50% threshold with correction for local background activity (A50%) seems optimal among the VOI methods evaluated.