@article {Lin665, author = {Christie Lin and Timothy Perk and Tyler Bradshaw and Jens Eickhoff and Michael Morris and John Humm and Steven Larson and Andrea Apolo and William Dahut and Scott Perlman and Glenn Liu and Robert Jeraj}, title = {An evaluation of reference tissue normalization in quantitative 18F-NaF PET/CT}, volume = {58}, number = {supplement 1}, pages = {665--665}, year = {2017}, publisher = {Society of Nuclear Medicine}, abstract = {665Objectives: Quantitative 18F-NaF PET/CT imaging has shown potential as a tool for identifying and monitoring malignant bone disease. However, natural fluctuations of 18F-NaF PET uptake within diseased and normal tissue have been observed. The aim of this study was to evaluate intra-patient variable uptake and evaluate the applicability of reference tissue normalization in quantitative 18F-NaF PET/CT.Methods: Thirty-three metastatic castrate-resistant prostate cancer patients received whole body test-retest scans prior to therapy. Our in-house Quantitative Total Bone Imaging software was used to identify, segment, and match NaF-avid bone lesions between scans using bone-specific SUV thresholds. Mean uptake was derived from whole healthy skeleton (SUVskel), normal bone regions (SUVbone), and normal tissue regions (SUVtissue). To assess if fluctuations in normal uptake correlated with fluctuations in diseased uptake, linear mixed effects models were used to model the relationship between relative mean difference (RMD) of normal uptake with lesion SUVs. RMD of lesion SUVs was then reassessed following normalization by normal uptake. Intra-patient repeatability was measured with the 95\% limits of agreement (LOA).Results: A total of 601 lesions were identified in the bone; both SUVmax (LOA: 1.01 [0.76, 1.35]) and SUVmean (LOA: 1.00 [0.87, 1.16]) were derived from individual lesions. RMD of SUVskel was not correlated to RMD of lesion SUVs. RMDs of SUVtissue, in regions such as in the liver and brain, were not correlated to RMD of lesion SUVs. RMDs of SUVbone, in regions such as the lumbar spine and pelvis, were similar across all regions. RMD of lesion SUVmean was not correlated to SUVbone. RMD of lesion SUVmax was correlated to RMD of SUVbone (P = 0.02); however, the magnitude of this correlation was small. When SUVbone was used as reference regions to normalize lesion SUVmax, measurement repeatability degraded (LOA: 1.00 [0.69, 1.45]).Conclusion: Variable tracer uptake across the whole healthy skeleton, within normal bone regions, and within normal tissue regions was poorly or not correlated to the variability in lesion uptake. Since reference tissue normalization is only useful when variations in reference region uptake and variations in diseased tissue uptake are linearly correlated, reference region normalization for quantitative 18F-NaF PET is not appropriate. Research Support: This study supported by Prostate Cancer Foundation Creativity Award, PCF Mazzone Challenge Award, and DOD PCRP Clinical Consortium Award, and conducted within the Prostate Cancer Clinical Trials Consortium.}, issn = {0161-5505}, URL = {https://jnm.snmjournals.org/content/58/supplement_1/665}, eprint = {https://jnm.snmjournals.org/content}, journal = {Journal of Nuclear Medicine} }