18F-FDG PET/CT Imaging Feedback for Tumor Voxel Dose-Response Assessment and Adaptive Dose Paining

Abstract

Conventional radiotherapy with uniform prescription dose can be safely used in clinics but has significant limitations to destroy local resistant tumor inches that cause treatment failures. Therefore, dose painting has been suggested to target the heterogeneity of tumor dose-response and improve radiotherapy precision. To implement dose painting correctly, the ability to identify and deliver optimum clinical doses safely and reliably to the resistant regions is essential. This dissertation aims to establish a framework of tumor voxel dose-response guided adaptive dose painting based on FDG-PET/CT imaging feedback for future clinical applications.

Serial FDG-PET/CT images were acquired during chemoradiotherapy from head and neck cancer patients. The changes of tumor voxel metabolic activity during treatment were quantified following the voxel-by-voxel deformable PET/CT image registration. Combining with the patient outcome, a tumor voxel control probability (TVCP) lookup table was created and used to guide treatment dose prescription for adaptive dose painting planning optimization. Additionally, the inter-/intra-tumoral variations of tumor voxel dose-response were quantified and their impacts on tumor control probability were evaluated. Finally, the effects of imaging/processing uncertainty on dose painting were determined by a simulation study, and the predictive capability of tumor voxel dose-response was assessed.

Tumor voxel dose-response distribution provides guidance to dynamically target tumor locoregional-resistant regions. The TVCP provides a quantitative objective to optimize the prescription dose distribution for individual tumors. Additionally, tumor voxel dose-response assessed using FDG-PET/CT images has a similar distribution to those of in vitro clonogenic assay. The inter/intra-tumoral variations cause large heterogeneity on the individual tumor controls and limit the efficacy of dose escalation/de-escalation using conventional uniform dose prescription. Finally, the adverse effects of PET imaging and processing uncertainty were highly dependent on individual tumor characteristics and should be managed individually using proper corrections. Tumor voxel dose-response can be reasonably predicted using 1 PET/CT feedback image acquired in the 3rd treatment week or 2 PET/CT feedback images acquired in the 2nd and 4th treatment week during chemoradiotherapy.