Noise-Induced Variability of Immuno-PET with Zirconium-89-Labeled Antibodies: an Analysis Based on Count-Reduced Clinical Images

Abstract

Purpose: Positron emission tomography (PET) with Zirconium-89 (Zr-89)-labeled antibodies can be used for in vivo quantification of antibody uptake. Knowledge about measurement variability is required to ensure correct interpretation. However, no clinical studies have been reported on measurement variability of Zr-89 immuno-PET. As variability due to low signal-to-noise is part of the total measurement variability, the aim of this study was to assess noise-induced variability of Zr-89 -immuno-PET using count-reduced clinical images. Procedures: Data were acquired from three previously reported clinical studies with [89Zr]antiCD20 (74 MBq, n = 7), [89Zr]antiEGFR (37 MBq, n = 7), and [89Zr]antiCD44 (37 MBq, n = 13), with imaging obtained 1 to 6 days post injection (D0–D6). Volumes of interest (VOIs) were manually delineated for liver, spleen, kidney, lung, brain, and tumor. For blood pool and bone marrow, fixed-size VOIs were used. Original PET list mode data were split and reconstructed, resulting in two count-reduced images at 50 % of the original injected dose (e.g., 37 MBq74inj). Repeatability coefficients (RC) were obtained from Bland-Altman analysis on standardized uptake values (SUV) derived from VOIs applied to these images. Results: The RC for the combined manually delineated organs for [89Zr] antiCD20 (37 MBq74inj) increased from D0 to D6 and was less than 6 % at all time points. Blood pool and bone marrow had higher RC, up to 43 % for 37 MBq74inj at D6. For tumor, the RC was up to 42 % for [89Zr]antiCD20 (37 MBq74inj). For [89Zr]antiCD20, (18 MBq74inj), [89Zr]antiEGFR (18 MBq37inj), and [89Zr]antiCD44 (18 MBq37inj), measurement variability was independent of the investigated antibody. Conclusions: Based on this study, noise-induced variability results in a RC for Zr-89-immuno-PET (37 MBq) around 6 % for manually delineated organs combined, increasing up to 43 % at D6 for blood pool and bone marrow, assuming similar biodistribution of antibodies. The signal-to-noise ratio leads to tumor RC up to 42 %.