Local Homogeneity of PET Images Visualizes Peri-Irradiation CD8 T Cell Exhaustion

Abstract

<h3>Purpose/Objective(s)</h3> The functional states of CD8⁺ T cells are associated with response to immune checkpoint inhibitors (ICI) and might become a potential biomarker indicating the timing of ICI. CD8⁺ T cells undergoing exhaustion (Tex) exhibit progressive loss of effector functions as well as metabolic insufficiency, such as suppressed glycolysis and restricted glucose uptake. We assumed that metabolic characteristics of Tex might be reflected by PET radiomic features, which could be used to noninvasively visualize the peri-irradiation dynamics of Tex and predict the optimal timing of ICI integration. Here, we aim to investigate the dynamics of Tex after ablative irradiation and develop an ¹⁸F-FDG PET/CT radiomic approach to assess tumor infiltrating Tex basing on non-small cell lung cancer (NSCLC) xenograft experiments. <h3>Materials/Methods</h3> C57BL/6 mice were inoculated with 1M Lewis Lung Carcinoma (LLC) cells subcutaneously on the right flank. On day 10 post-tumor inoculation, 20Gy was delivered to the tumors in a single fraction. For the control (D0) and those irradiated (D3, D7, D14 and D21 post-irradiation), mice received micro PET-CT scans and were sacrificed immediately after scanning, with tumors and spleens collected. The abundance of tumor-infiltrating immune cell subsets were estimated by deconvoluting the tumor microarray data with CIBERSORTx. ICI-responsive early Tex (CD8⁺PD-1^intTcf-1⁺) and ICI-nonresponsive terminal Tex (CD8⁺PD-1^hiTIM-3⁺TOX⁺) were identified by flow cytometry. Tumors on PET/CT images were manually contoured, and radiomic features were extracted by a compact proton therapy machine. Robust features which passed Test-retest analysis will undergo further selection. Least absolute shrinkage and selection operator (LASSO) logistic regression was used to select radiomic features significantly correlated with the infiltration of early or terminal Tex. <h3>Results</h3> After ablative irradiation, the tumor CD8⁺ T cell infiltration increased from D3 through the peak on D14, and then declined sharply to a below-baseline level on D21. However, dysfunctional terminal Tex significantly predominated CD8⁺ T cells from D7 to D14 (69.44%∼ 76.35%), indicating a potential window for ICI treatment between D3-D7 post-irradiation. 40 (out of 57) CT and 9 (out of 61) PET radiomic features of good robustness were brought into LASSO regression. Gray Level Co-occurrence Matrix (GLCM)-inverse variance and Gray Level Run Length Matrix (GLRLM)-long run emphasis on PET images were found to be inversely correlated with the abundance of terminal Tex, showing more heterogeneous local distribution of FDG uptake correlated with higher T exhaustion. <h3>Conclusion</h3> The dynamics of Tex help identify the proper timing to combine ICI with irradiation. Local homogeneity of PET images may provide a noninvasive approach to visualize peri-irradiation T cell exhaustion, and exhibit the potential to estimate ICI timing among more radiation dose-fractionation schedules in future study.