Repeatability of Quantitative ¹⁸F-NaF PET: A Multicenter Study

Abstract

Purpose: ¹⁸F-NaF, a positron emission tomography (PET) radiotracer of bone turnover, has shown potential as an imaging biomarker for assessing therapeutic response of bone metastases. This study aims to evaluate the repeatability of ¹⁸F-NaF PET-derived SUV metrics in individual bone lesions from patients in a multicenter study. Methods: Thirty-five metastatic castrate-resistant prostate cancer patients with multiple metastases received two whole-body (test-retest) ¹⁸F-NaF PET/computed tomography (CT) scans 3±2 days apart from one of three imaging sites. A total of 411 bone lesions larger than 1.5 cm³ were automatically segmented using standardized uptake value (SUV) > 15 g/mL threshold. Two levels of analysis were performed: lesion-level, where measures were extracted from individual lesion regions of interest (ROI), and patient-level, where all lesions within a patient were grouped into a single ROI for analysis. Uptake was quantified with the maximum (SUV_max), average (SUV_mean), and total (SUV_total) SUV. Test-retest repeatability was assessed using Bland-Altman analysis, intraclass correlation coefficient (ICC), coefficient of variation (CV), critical percent difference, and repeatability coefficient. The 95% limits of agreement (LOA_95%) of the ratio between test-retest measurements were calculated. Results: At the lesion-level, CV for SUV_max, SUV_mean, and SUV_total were 14.1%, 6.6%, and 25.5%, respectively. At the patient-level, CV was slightly smaller: 12.0%, 5.3%, and 16.5%, respectively. ICC was excellent (ICC>0.95) for all imaging metrics. Lesion-level LOA_95% for SUV_max, SUV_mean, and SUV_total were (0.76, 1.32), (0.88, 1.14), and (0.63, 1.71), respectively. Patient-level LOA_95% were slightly narrower at (0.79, 1.26), (0.89, 1.10), and (0.70, 1.44), respectively. We observed significant differences in the variance and sample mean of lesion-level and patient-level measurements between imaging sites. Conclusion: Repeatability of ¹⁸F-NaF PET/CT of SUV_max, SUV_mean, and SUV_total were similar between lesion- and patient-level regions of interest. We found significant differences in lesion-level and patient-level difference distributions between sites. These results can be used to establish ¹⁸F-NaF PET-based treatment response assessment criteria at the lesion- and patient-levels. ¹⁸F-NaF PET demonstrates repeatability levels useful for clinical quantification of bone lesion response to therapy.