Impact of Different Standardized Uptake Value Measures on PET-Based Quantification of Treatment Response

Abstract

PET-based treatment response studies typically measure the change in the standardized uptake value (SUV) to quantify response. The relative changes of different SUV measures, such as maximum, peak, mean, or total SUVs (SUV(max), SUV(peak), SUV(mean), or SUV(total), respectively), are used across the literature to classify patients into response categories, with quantitative thresholds separating the different categories. We investigated the impact of different SUV measures on the quantification and classification of PET-based treatment response. Sixteen patients with solid malignancies were treated with a multitargeted receptor tyrosine kinase inhibitor, resulting in a variety of responses. Using the cellular proliferation marker 3'-deoxy-3'-(18)F-fluorothymidine ((18)F-FLT), we acquired whole-body PET/CT scans at baseline, during treatment, and after treatment. The highest (18)F-FLT uptake lesions (~2/patient) were segmented on PET images. Tumor PET response was assessed via the relative change in SUV(max), SUV(peak), SUV(mean), and SUV(total), thereby yielding 4 different responses for each tumor at mid- and posttreatment. For each SUV measure, a population average PET response was determined over all tumors. Standard deviation (SD) and range were used to quantify variation of PET response within individual tumors and population averages. Different SUV measures resulted in substantial variation of individual tumor PET response assessments (average SD, 20%; average range, 40%). The most extreme variation between 4 PET response measures was 90% in individual tumors. Classification of tumor PET response depended strongly on the SUV measure, because different SUV measures resulted in conflicting categorizations of PET response (ambiguous treatment response assessment) in more than 80% of tumors. Variation of the population average PET response was considerably smaller (average SD, 7%; average range, 16%), and this variation was not statistically significant. Differences in tumor PET response were greatest between SUV(mean) and SUV(total) and smallest between SUV(max) and SUV(peak). Variations of tumor PET response at midtreatment and posttreatment were similar. Quantification and classification of PET-based treatment response in individual patients were strongly affected by the SUV measure used to assess response. This substantial uncertainty in individual patient PET response was present despite the concurrent robustness of the population average PET response. Given the ambiguity of individual patient PET responses, selection of PET-based treatment response measures and their associated thresholds should be carefully optimized.