Natural History and Clinical/Dosimetric Determinants of Osteoradionecrosis in a Large Cohort of Head and Neck Cancer Following Curative Radiotherapy: Debunking the Myth of Decreased Rates of Osteoradionecrosis in the Modern Radiotherapy Era

PURPOSE/OBJECTIVE(S)

Osteoradionecrosis (ORN) of the mandible is the most devastating toxicity following head and neck cancer (HNC) radiotherapy (RT). The rate of ORN occurrence has been debatable in the modern era of advanced RT. We aim to determine the natural history and time to ORN development in a large cohort of HNC.

MATERIALS/METHODS

After IRB approval, we identified HNC patients treated with curative-intent RT between 2005 and 2020 at MD Anderson Cancer Center. Dental oncology notes were reviewed and all dental procedures were recorded. Available dose volume histograms (DVHs) for the segmented mandibular volumes were extracted. Based on our previously published ORN normal tissue complication probability (NTCP) model, any dental procedure pre-RT is a clinical high-risk factor and, therefore, data were coded as (high vs. low clinical risk) accordingly. We also coded the dosimetric risk factors to (high vs. low dosimetric risk) according to our published DVH thresholds (high dosimetric risk if any applies: V45≥40%, V55≥25%, and/or D30 ≥40 Gy). Finally, patients were classified to four groups according to combined clinical and dosimetric risk factors (No, clinical, dosimetric, and both). We used the Kaplan-Meier method to calculate time to ORN development and ORN-free survival. For time to ORN development, any grade ORN occurrence was coded as event and all others were censored. For ORN-free survival, death and ORN were coded as events and all others were censored. Log-rank test was used to compared curves of different risk groups.

RESULTS

One thousand eight hundred sixty-six patients were included. Median follow-up was 38 months (range 4-162). ORN was reported in 252 patients (13.5%). The median time to ORN development was 18.5 months (range 4-145). 95 patients (37.7%) developed ORN after 2-years post-RT. The 1-, 3-, and 5-year ORN rates were 4.7%, 12.7%, and 17.8%, respectively. The 1-, 3-, and 5-year ORN-free survival were 94.3%, 85.5%, and 80.3%, respectively. There were statistically significant differences (P < 0.0001) between ORN-free survival in different clinical/dosimetric risk factors. The patients with no, clinical-only, dosimetric-only, and both clinical and dosimetric risk factors were 35%, 19%, 22%, and 24%, respectively. The 5-year ORN-free survival was 94.3%, 89.8%, 76.3%, and 69.6% for patients with no, clinical-only, dosimetric-only, and both clinical and dosimetric risk factors, respectively. The hazard-ratio (HR) of ORN development in clinical-only, dosimetric-only, and both clinical and dosimetric risk groups was 2.1, 5.4, and 7.5 compared to the no risk group (P<0.05 for all).

CONCLUSION

Our findings indicate that ORN remains a remarkable toxicity hazard for HNC survivors. A prolonged surveillance time is required for the majority of HNC survivor since more than one-third of the ORN events occurred after 2-year follow-up. Patients with combined clinical and dosimetric risk factors have a staggering ORN risk profile and are proper candidates for future prophylactic pharmacotherapy clinical studies.

Deep Learning Based Prognostic Prediction in Oropharyngeal Cancer Patients Using Multiparametric MRI Inputs

PURPOSE/OBJECTIVE(S)

While prognostic outcomes for oropharyngeal cancer (OPC) patients have improved in recent years, patients still face a non-negligible risk of disease recurrence or death. Accurately predicting post-therapy prognosis would be highly valuable for risk stratification and treatment guidance for OPC patients. Recent studies using PET/CT data have demonstrated the effectiveness of large-scale, end-to-end image-based deep learning (DL) models for predicting progression-free survival (PFS) in OPC patients. Multiparametric MRI (mpMRI), which combines anatomical and functional MRI sequences, has the potential to offer similar results, and has the added advantage of high-frequency longitudinal imaging capabilities, such as through MR-Linac devices. Therefore, this study aimed to develop a DL model using mpMRI data to predict PFS, and to evaluate the impact of anatomical and functional input channels on model performance.

MATERIALS/METHODS

From a large-scale head and neck cancer database at MD Anderson Cancer Center, treatment-naïve OPC patients with available pre-radiotherapy mpMRI imaging were selected for this study. mpMRI images used for this study included T2-weighted images (T2) and apparent diffusion coefficient (ADC) maps. PFS event status was defined as having either a local, regional, or distant failure, and/or death; data were right censored if an event had not occurred. Images were resampled to the T2 resolution, normalized to a [-1,1] scale, and cropped to the field of view of the ADC image for use in DL models. A DL convolutional neural network model based on the DenseNet121 architecture from the Medical Open Network for AI (MONAI) Python package using a negative log-likelihood loss function was implemented. The model used mpMRI images as input channels and 20 output channels representing the different time intervals of the predicted PFS conditional probabilities of surviving that time interval; final PFS in days was obtained by summing the cumulative probability of surviving each interval times the interval duration. A 5-fold cross validation approach was used for model training and evaluation. Separate models using only T2, only ADC, and T2 + ADC channel inputs were compared. Model performance was measured using the C-index.

RESULTS

Out of 1154 patients, 404 met inclusion criteria. The overall PFS event rate was 16%. Median C-index values from the 5-fold cross validation were 0.62, 0.67, and 0.69 for the ADC, T2, and T2+ADC models, respectively.

CONCLUSION

Using large-scale datasets and open-source DL implementations, we find that OPC PFS prediction models using mpMRI data yield modest but comparable performance to existing models (i.e., state-of-the-art reference performance using PET/CT). Moreover, combining mpMRI channels may increase the performance of models for OPC prognostic prediction. Future work will involve integration of additional timepoints, additional mpMRI images, clinical variables, and saliency maps.