2838 - Can Pre-Radiation Therapy MRI Imaging Predict Osteoradionecrosis in Oral Cavity Cancer?

Purpose/Objective(s): Osteoradionecrosis (ORN) is a well-known side effect after intensity modulated radiation therapy (IMRT) to the head and neck. We aim to determine if radiographic imaging markers pre-radiation therapy (RT) has any role in predicting development of ORN.

Materials/Methods: This retrospective study analyzed oral cavity cancer patients treated with definitive or adjuvant radiation at a single institution from 2014-2023. Patients without pre-RT MRI imaging, or those receiving reirradiation or brachytherapy were excluded. MRI radiomarkers, including enhancement on post contrast T1 and edema seen as hyperintense T2 STIR signal, and Time Intensity Curves (TIC) were collected. TICs were measured on pre-RT dynamic contrast-enhanced (DCE) MR series on the ipsilateral mandible (region of interest, ROI) and classified as Type 1-5: Type 1 with no contrast uptake resulting in a horizontal type curve, type 2 with progressive enhancement, type 3 with a delayed plateau, type 4 with a delayed washout, and type 5 with fast initial contrast uptake followed by progressive late enhancement. Dental exams before RT were recorded as normal or abnormal. Univariate analysis was performed using Chi-squared tests, Fisher’s exact test, and logistic regression to compare patients with and without ORN.

Results: The cohort included 59 patients with a median follow-up of 18.3 months, with 27% (n=16) developing ORN. Median dose was 66 Gy (50 Gy – 74.70 Gy) in all patients and 66Gy (60Gy-69.96 Gy) in ORN patients. All ORN cases occurred within the 60 Gy isodose line. 47% (n=27) and 64% (n=35) of patients had enhancement or mild/obvious T2 edema present, respectively. 43% (n=17), 18% (n=7), 15% (n=6), 2.5% (n=1), and 23% (n=5) of patients had TIC Type 1, 2, 3, 4, and 5, respectively. ORN patients were significantly more likely to exhibit T1 enhancement compared to those without ORN (80% vs. 35%, p=0.003). Mild/obvious T2 edema was present in 93% of ORN vs. 54% of non-ORN patients (p=0.008). TIC Type 1 was observed in 55% of non-ORN patients but was absent in ORN patients, whereas, all ORN patients had TIC Type 2 or higher (100% vs. 45%, p=0.005). TIC Type 3 or higher was strongly associated with ORN (OR: 23.0, 95% CI: 3.47–463, p=0.006). Among ORN cases, 78% occurred within 1 cm of the ROI (average 10 mm²), and 91% were within areas of enhancement. TIC Type 1 demonstrated high sensitivity (89%) and strong negative predictive value (96%), making it effective for ruling out ORN risk. Although abnormal dental exams were more frequent in ORN patients, 44% of ORN vs 28% non-ORN patients, this difference was not statistically significant (p=0.264).

Conclusion: MR radiomarkers on pre-RT MRI such as T1 enhancement, T2 STIR hyperintensity, and TIC of Type 2 or higher can help predict patients that may be at a higher risk for developing ORN, while having a TIC Type 1 may be a strong negative predictor for developing ORN. The majority of MRI signal abnormality were within the high-risk target volume and unlikely to be a robust avoidance area in treatment planning.