Patterns of response in head and neck cancer subregions using daily Quantitative MRI from MR-guided radiation therapy

Quantitative MRI (qMRI) is an emerging technique for characterizing tissue microenvironments. Prior work has shown that daily qMRI during MR-guided radiotherapy (MRgRT) detects differential responses to hypofractionated radiotherapy (RT) in head and neck squamous cell carcinoma (HNSCC) within the primary gross tumor volume (GTVp). This exploratory study investigates whether subregions of HNSCC tumors show distinct qMRI responses to hypofractionated RT using contoured GTVp subregions and voxelwise data.

Eighteen subjects with advanced HNSCC underwent MRgRT with daily qMRI over 15 fractions (total dose 50, 55, or 60 Gy). Daily intravoxel incoherent motion (IVIM) and relaxometry sequences were acquired on a 1.5 T MR-Linac. Median ADC, D, D*, f, T₁, and T₂ were calculated for concentric GTVp subregions and an intra-subject control contour in paraspinal muscle. Mixed effect models were fit for each parameter. Voxelwise analysis included quantile tracking, histograms, and paired tests over the length of RT.

Significant changes (p < 0.05) were observed in ADC, D, f, T₂, and T₁ within GTVp subregions. ADC, D, f, and T₂ changes were uniform, while T₁ varied spatially. Spatial patterns were detected in ADC and T₁, with marginal effects in D, f, and T₂. Voxelwise analysis revealed significant changes in all parameters except D* after correction. ADC, D, and f increased consistently; T₁ and T₂ changes were quantile dependent. These trends may reflect biological processes such as declining cellularity, reperfusion, or edema.

In this exploratory study, qMRI detected spatial heterogeneity in HNSCC. These findings support its potential role in guiding biologically adaptive RT strategies.