RT Journal Article
SR Electronic
T1 Mr.
JF Journal of Nuclear Medicine
JO J Nucl Med
FD Society of Nuclear Medicine
SP 1202
OP 1202
VO 60
IS supplement 1
A1 Abhinay Joshi
A1 Deepak Behera
A1 Ian Wilson
A1 William Le
A1 John Sunderland
A1 Bonnie Clarke
A1 Tessa Jhonson
YR 2019
UL http://jnm.snmjournals.org/content/60/supplement_1/1202.abstract
AB 1202Objectives: Harmonization of image acquisition and reconstruction parameters are essential in a multi-center PET imaging clinical trial. ImaginAb has partnered with CTN to validate PET scanners and harmonize image acquisition and reconstruction parameters for a multicenter PET imaging trial using 89Zr tracers. Methods: For standardization, all sites calibrated their dose calibrators for 89Zr to match activity defined by the 89Zr manufacturer. An anthropomorphic chest phantom with lung fields and 11 spherical objects with inner diameters ranging from 7 to 35 mm reproducibly secured at specific locations within the phantom was used. 1.0 mCi of 89Zr-oxalate combined with DTPA for chemical stability into background of the phantom. For the 11 spherical lesions, 1.0 mCi of similarly stabilized 89Zr was added to a 1.00 liter of water, creating a 9.5:1 lesion to background ratio. The phantom is filled on-site. Activities and contrasts for the phantom fill were chosen to quantitatively mimic the pharmacological concentrations in tumor and normal tissue from the biodistribution studies of a 89Zr labeled CD8 T-cell tracer, 89Zr-IAB22M2C. To allow for use of quantitative endpoints in future trials, prospectively harmonized reconstructions were chosen for each of the scanner models. Eighteen phantom images from nine sites (twenty sites planned) were acquired. 4 minute/bed scans were used to parallel noise properties of the human imaging protocol. 10 minute/bed scans were used for a more robust, less noisy quantitative assessment. All images were reconstructed using parameters specific for a scanner model (ranging from 2i/16s to 4i/16s) with varied post-reconstruction gaussian filters 4 mm-6.4 mm. Phantom image evaluations were performed by CTN. The qualitative evaluation includes a review of artifacts, PET to CT alignment, and protocol adherence. Quantitative assessments include calibration accuracy, recovery coefficient curves, coefficient of variance (COV) and contrast to noise ratio (C/N). For scanner calibration an average SUV of thirty regions of interest (ROI) from homogenous areas of the image were used. A mean COV using thirty ROIs was used for assessment of COV. Contrast to noise ratio was calculated using standard NEMA methods. Results: 89Zr oxalate was found to crash out of solution in the phantoms unless stabilized by chelation with DTPA. Calibration of dose calibrators to a common source was necessary, with gain changes of >10% common. Notably, scanner calibrations for 89Zr were accurate across all vendors. All sites except one passed the qualitative evaluations. For the failed scanner, the phantom demonstrated remarkable image non-uniformities on the order of 20-30% from center to periphery. Calibration error was 23%, 13% greater than acceptance threshold of ±10%, further supporting the qualitative evaluation of the failed phantom. As expected, spheres of smaller size gave higher C/N than larger spheres irrespective of the scanner model. Conclusions: A first ever endeavor of 89Zr-based multi-center scanner validation program was successfully launched. Stability of 89Zr with DTPA was necessary, as was calibration of dose calibrators to a common standard. Overall, scanners across vendors demonstrated excellent calibration accuracy with 89Zr. COV, C/N and calibration measures from 4 min and 10 min acquisitions didn’t vary significantly. Ongoing work includes validating the remainder of the twenty sites, gaining a deeper understanding of variability from different scanner models, and applying mitigation measures to achieve a network of sites with harmonized PET scanners for future clinical trials using 89Zr PET tracers. $$graphic_02ECA0CD-12A9-44B5-831C-62905E0A9A48$$