Comparison of six quantitative methods for the measurement of bone turnover at the hip and lumbar spine using 18F-fluoride PET-CT

Abstract

The aim of this study was to evaluate the relationship between different quantification methods used for the measurement of bone plasma clearance (K(i)) using F-PET at the hip and lumbar spine. Twelve healthy postmenopausal women aged 52-71 years were recruited. Each participant underwent 60-min dynamic F-PET scans at the lumbar spine and hip on two separate occasions with an injected activity of 90 and 180 MBq, respectively. Image-derived input functions were obtained at the aorta from the lumbar spine scans. K(i) was evaluated using a three-compartment four-parameter model (K(i-4k)), three-compartment three-parameter model (K(i-3k)), Patlak analysis (K(i-Pat)), spectral analysis (K(i-Spec)) and deconvolution (K(i-Decon)). Standardized uptake values (SUVs) were also measured. The Pearson correlation between K(i-4k) and K(i-3k), K(i-Pat), K(i-Spec), K(i-Decon) and SUV were 0.91, 0.97, 0.94, 0.95 and 0.93, respectively, with a significance of P less than 0.0001. The differences between the correlations measured using Fisher's Z-test were not significant (P>0.05). Bland-Altman analysis showed that the limits of agreement for K(i) measured as the SD of the differences were 0.0082 (25.9%), 0.0062 (11.7%), 0.0098 (20.1%) and 0.0056 (25.5%) ml/min/ml, respectively, and the biases were -0.0081 (-23.8%), -0.0075 (-23.7%), -0.0107 (-29.5%) and -0.0015 (0.8%) ml/min/ml, respectively. All five methods of quantification (K(i-3k), K(i-Pat), K(i-Spec), K(i-Decon) and SUV) strongly correlated with K(i-4k). Although systematic differences of up to 29% were found between K(i-4k) and the other methods (K(i-3k), K(i-Pat), K(i-Spec) and K(i-Decon)), these should not affect the conclusions of clinical studies, provided the methods are applied consistently. However, care should be taken when comparing reports that use different methods of quantification.