We assessed the precision of lumbar spine (18)F-PET measurements based on 58 scans performed on 20 postmenopausal women. The percentage coefficient of variation (%CV) (95% confidence interval) was 9.2% (7.5-11.8) for standardised uptake values, 11.7% (9.5-14.9) for plasma clearance measurements using the Patlak method and 14.5% (11.7-18.5) for plasma clearance measurements using the Hawkins three-compartment model. (18)F-Fluoride positron emission tomography ((18)F-PET) is a non-invasive technique that allows the assessment of regional bone turnover in patients with metabolic bone disease. Knowledge of the precision errors of (18)F-PET measurements is important for planning the number of subjects required for research studies. Twenty osteoporotic postmenopausal women had (18)F-PET scans of the lumbar spine at 0, 6 and 12 months after stopping long-term bisphosphonate treatment. No significant changes in the PET measurements were seen over the 12-month period, and the data were deemed suitable for a precision study. Precision errors were evaluated for standardised uptake values (SUVs) and for the fluoride plasma clearance to bone mineral (K (i)) determined using the Patlak and Hawkins methods. Precision errors were expressed as the %CV and were calculated for the mean L1-L4 region and for individual vertebrae. %CV (95% confidence interval) for the L1-L4 region was 9.2% (7.5-11.8) for SUV, 11.7% (9.5-14.9) for K (i) measured using the Patlak method and 14.5% (11.7-18.5) for K (i) measured using the Hawkins method. There was no significant difference between precision errors obtained for the L1-L4 region and those obtained for a single vertebra. SUV measurements showed the smallest precision error followed by the Patlak method, while the Hawkins method gave the largest error. Measuring a smaller region of interest did not increase the precision error, suggesting that the factor determining the errors may be scanner calibration.
Read full abstract