Harmonization of Quantitative Values in Bone SPECT/CT for Medication-Related Osteonecrosis of the Jaw.

Abstract

The increased use of antiresorptive and antiangiogenic agents in patients with osteoporosis and cancer is closely linked to decreases in quality of life attributable to medication-related osteonecrosis of the jaw (MRONJ), emphasizing the need for accurate diagnosis. Recent advances in quantitative SPECT/CT (Q-SPECT/CT) using bone scintigraphy have enhanced its utility for the early detection and staging of MRONJ. However, the lack of harmonization criteria for SUVs across various scanners hampers multicenter studies. Methods: To address this issue, we conducted a multicenter phantom study to identify harmonization criteria for Q-SPECT/CT in MRONJ and evaluate a software-based harmonization approach. A custom-made brain tumor phantom with 6 hot spheres was filled with a ^99mTc-hydroxymethylene diphosphonate solution (17.5 kBq/mL in the background and 70 kBq/mL in the hot spheres, yielding a hot sphere-to-background ratio of 4). SPECT/CT was performed at 7 institutions using 8 scanners in accordance with clinical protocols. A cylindric phantom was used to calculate the becquerel calibration factor, and a tool for analyzing region of interest and volume of interest was used for SUV conversion, analysis, and harmonization. The harmonization criterion was the median SUV_max ± 30% across scanners. If the SUV_max exceeded this range, then a 3-dimensional gaussian filter was applied. Results: The SUV_mean in the background region remained within 0.95-1.05 (relative to the theoretic value of 1.00) both before and after harmonization, with minimal interscanner differences. Conversely, for the hot spheres, the maximum coefficients of variation for SUV_max, SUV_peak, and SUV_mean improved from 122%, 68%, and 71%, respectively, before harmonization to 48%, 35%, and 37%, respectively, after harmonization. Conclusion: We identified a harmonization criterion for Q-SPECT/CT in MRONJ and demonstrated that our software-based approach effectively reduces interscanner variability without compromising clinical image quality or requiring additional image reconstruction, supporting its utility in multicenter clinical studies.