We evaluated acute toxicity and disease control in 42 patients receiving thoracic radiotherapy with concurrent KRAS G12C inhibitors. Toxicity included 13 grade 2 events (31%) and one grade 3 pneumonitis. There were no significant differences in toxicity between patients who held versus continued inhibitors during radiotherapy. The estimated cumulative incidence of local failure at 6 months was 5%.
Purpose/Objective(s) Proton beam therapy (PBRT) has emerged as a treatment strategy for head and neck cancers. However, there is an ongoing effort to understand normal tissue effects with protons, owing to their biologic uncertainties. Osteoradionecrosis (ORN) is a known late toxicity of radiation therapy in the treatment of head and neck cancers. We hypothesized that ORN risk can be mitigated by careful treatment planning, accounting for both linear energy transfer (LET) and dose. We set out to characterize the interactions between LET, dose, and ORN risk. Materials/Methods We retrospectively identified 6 patients treated definitively with PBRT and concurrent chemotherapy for HPV+ oropharyngeal squamous cell carcinoma who developed ORN and performed a detailed dosimetric analysis. Patient, disease, and treatment characteristics were collected. Target ORN contours were generated using consensus information from imaging (MRI and/or CT) and dental examination findings at the time of ORN diagnosis. A control region contour was defined as the overlap between mandible and the 20% isodose line excluding the ORN region and with a 5 mm expansion. Voxel level dose (Gy) and LET (MeV/cm) data were extracted for both contours. Dose was split into 1.5 Gy bins (range 3–36 Gy), and the mean ± SD LET was plotted for ORN and control voxels against dose. From these plots, LET > 4 KeV/µm was selected as a high-LET threshold beyond which increasing LET contributed appreciably to ORN. A generalized linear mixed-effects model (GLMM) was fit with centered dose (dose minus mean) and a dichotomized LET variable (≤4 vs >4 KeV/µm). Results All 6 patients were males with stage I-III base of tongue and/or tonsillar SCC treated with PBRT to 70 CGE in 35 fractions (30 CGE elective dose) with concurrent high dose cisplatin. One patient had induction carboplatin/paclitaxel. Mean time from RT completion to ORN diagnosis was 18.3 months (range 2-34 months). All ORN was diagnosed in the posterolateral mandible. A total of 23,609 control voxels and 758 ORN voxels were analyzed. Control voxels had a mean dose of 20.4 Gy (median 21.6 CGE, IQR 15.9 CGE; range 3.11–36.1 CGE), whereas ORN voxels had a higher mean dose of 28.5 Gy (median 30.1 CGE, IQR 3.04 CGE; range 7.05–33.5 CGE). Binned dose–LET plots demonstrated that mean LET was consistently higher in ORN voxels across overlapping dose bins, with LET exceeding 4 KeV/µm in the highest-risk voxels. At LET ≤ 4, the GLMM indicated that increasing dose above the mean substantially increased ORN risk (OR 1.29) with voxels at mean dose and high LET having higher odds of ORN (OR 5.62). However, at LET > 4, ORN probability is already elevated and the incremental effect of increasing dose is reduced (OR 0.96). Conclusion In this cohort of patients with ORN following PBRT for OPSCC, an LET of > 4 KeV/µm seemed to drive ORN even at moderate doses, whereas at an LET ≤ 4 KeV/µm, dose was more influential. Models to predict ORN based on LET and dose are in process. Our data suggest that LET-based PBRT planning is needed to minimize ORN risk.
