Abstract 1794: Lymphocyte apoptosis as a predictive biomarker for radiotherapy de-intensification in EBV-associated nasopharynx cancer

Aims: Emerging evidence in viral-associated oro- and naso-pharynx cancers supports the concept of radiotherapy (RT) dose de-intensification in these radiosensitive tumours. This notion is particularly relevant in the background that a majority of head and neck cancer patients develop severe late toxicities despite modern precision RT. A reasonable strategy could thus entail stratifying patients at risk of late toxicities to dose de-intensification. Here, we investigated the utility of a lymphocyte apoptosis (RILA) assay to predict for late toxicities in EBV-positive nasopharynx cancer patients. We also test if a multi-modal approach incorporating DNA damage induction and repair improves the prediction of clinical radiosensitivity. Methods: Assays were assessed retrospectively among survivors of a randomised controlled trial (NCC-0902), where patients were assigned to weekly cisplatin-RT with or without neoadjuvant gemcitabine, carboplatin, and paclitaxel. Late toxicities were assessed by CTCAE v.2 at the following intervals - 2-mo year 1, 4-mo year 2, 6-mo years 3-5, and annually thereafter; and considered severe if CTCAE v.2 ≥Grade 2. RILA was based on a fluorogenic inhibitor of caspases assay (48 h post-8 Gy). DNA damage induction and repair were assessed by semi-automated scoring of ϒH2AX foci at 30 min post-1 Gy and 24 h post-4 Gy, respectively. Results: Median follow-up of patients was 5.7 (range 4.6-7.4) years. Clinical and treatment indices, including assigned study arm and RT dosimetry to normal tissue structures, were balanced between patients with and without late toxicities (p >0.05). We observed a trend between decreasing RILA scores and increased risks of late toxicities; odds ratio (OR) 1.27 (95% CI = 0.92-1.72) for lowest quartile cut-off, corresponding to an AUC of 0.608 for prediction accuracy (p = 0.049). A multi-modal approach incorporating DNA damage induction and repair did not improve upon the predictive accuracy of RILA for clinical radiosensitivity. Independently, DNA damage induction and repair were not associated with risks of late toxicities; OR 1.03 (95% CI = 0.69-1.53, p = 0.901) and 1.04 (95% CI = 0.73-1.49, p = 0.829), respectively. Interestingly, onset of late toxicities was correlated to an improved disease-free survival in our cohort (5-y DFS = 94.1% vs 73.3%, no late toxicity, p = 0.003), which suggests a common mechanism underlying tumour and normal tissue radiosensitivity in EBV-positive nasopharynx cancer. Conclusions: