Garcia, Darwin A

Abstract Ataxia Telangiectasia–mutated (ATM) inhibitors are being developed as radiosensitizers to improve the antitumor effects of radiotherapy, but ATM inhibition can also radiosensitize normal tissues. Therefore, understanding the elevated risk of normal tissue toxicities is critical for radiosensitizer development. This study focused on modeling the relationship between acute mucosal toxicity, radiation dose, fractionation schedule, and radiosensitizer exposure. The ATM inhibitor WSD0628 was combined with single or fractionated doses of radiation delivered to the oral cavity or esophagus of Friend Leukemia virus B (FVB) mice. The potentiation by WSD0628 was quantified by a sensitizer enhancement ratio (SER), which describes the changes in radiation tolerance for radiation combined with WSD0628 relative to radiation-only regimens. WSD0628 profoundly enhanced radiation-induced acute oral and esophageal toxicities. For oral mucosal toxicity, the enhancement by WSD0628 with 3 fractions of radiation resulted in an SER ranging from 1.3 (0.25 mg/kg) to 3.1 (7.5 mg/kg). For the 7.5 mg/kg combination, the SER increased with increasing number of fractions from 2.2 (1 fraction) to 4.3 (7 fractions) for oral toxicity and from 2.2 (1 fraction) to 3.6 (3 fractions) for esophageal toxicity, which reflects a loss of the normal tissue sparing benefit of fractionated radiation. These findings were used to develop a modified biologically effective dose model to determine alternative radiation schedules with or without WSD0628 that result in similar levels of toxicity. Successful radiosensitizer dose escalation to a maximally effective therapeutic dose will require careful deliberation of tumor site and reduction of radiation dose volume limits for organs at risk.

In this study, we investigated whether radiomics features from pre-treatment positron emission tomography (PET) images could be used to predict disease progression in patients with HPV-positive oropharyngeal cancer treated with definitive proton or x-ray radiotherapy. Machine learning models were built using a dataset from Mayo Clinic, Rochester, Minnesota (n = 72) and tested on a dataset from Mayo Clinic, Phoenix, Arizona (n = 22). A total of 71 clinical and radiomics features were considered. The Mann–Whitney U test was used to identify the top 2 clinical and top 20 radiomics features that were significantly different between progression and progression-free patients. Two dimensionality reduction methods were used to define two feature sets (manually filtered or machine-driven). A forward feature selection scheme was conducted on each feature set to build models of increased complexity (number of input features from 1 to 6) and evaluate model robustness and overfitting. The machine-driven features had superior performance and were less prone to overfitting compared to the manually filtered features. The four-variable Gaussian Naïve Bayes model using the ‘Radiation Type’ clinical feature and three machine-driven features achieved a training accuracy of 79% and testing accuracy of 77%. These results demonstrate that radiomics features can provide risk stratification beyond HPV-status to formulate individualized treatment and follow-up strategies.