Kinetic Modeling and Radiomics Analysis of the Heterogeneity of Tumor-Associated Macrophages and Metabolism in Head and Neck Cancer

Purpose: Tumor-associated macrophages (TAMs) play an important role in the initiation and progression of head and neck squamous cell carcinoma (HNSCC). Herein, we evaluated the tumor tissue heterogeneity uptake of a CD163+ PET tracer, 64CuICT-01, and compared it with 18FFDG using kinetic modeling and radiomics analysis in a mouse oral carcinoma 1 (MOC1) HNSCC mouse model. Methods: MOC1 cells were implanted on the flanks of four C57BL/6 mice. At 3 weeks post tumor implantation, these mice underwent a 60 min dynamic scan post-injection of 18FFDG and then another 60 min dynamic scan the next day post-injection of 64CuICT-01. Nine tumor volumes were delineated using a region-growing algorithm, which were further divided into tissue volumes based on tracer uptake using k-means clustering. Kinetic modeling was performed using compartmental models and graphical Logan/Patlak. Standardized uptake value (SUV) PET images were generated using time windows of 1–10 min and 45- 60 min. A total of 91 textural radiomic features were obtained using Pyradiomics, and key features were selected using the coefficient of variance (CV). Results: k-means clustering identified three distinct tissue uptakes for 64CuICT-01, and only two tissues with high and low uptakes for 18FFDG. The 2TC4K and Logan models were found to be suitable models for modeling the kinetics of 64CuICT-01. 18FFDG generally yielded greater variances in quantitative parameters across tissues within the tumors than those of 64CuICT-01, except for K1. However, greater variances in radiomic features were obtained for 64CuICT-01 than for 18FFDG. 64CuICT-01 also yielded six more features with CV > 0.7 compared to 18FFDG. Conclusion: 64CuICT-01 showed greater differences in uptake kinetics and textural information, while 18FFDG provided greater differences in quantitative values across the tissues within the tumor. 64CuICT-01 and 18FFDG showed the heterogeneous nature of the tumor microenvironment, and their combination would enable us to understand the tumor biology and hence improve tumor diagnosis.