Regional recurrence of oropharyngeal cancer after definitive radiotherapy : a case control study

Background: Elective treatment of lymph nodes in oropharyngeal cancer (OPC) has impact on both regional recurrences (RR) and risk of late side effects. This study was performed to quantify the dose-dependent impact on RR and overall survival (OS) in a prospectively collected cohort of OPC from the ARTSCAN study with emphasis on elective treatment. Methods: ARTSCAN is a previously published prospective, randomized, multicentre study of altered radiotherapy (RT) fractionation in head and neck cancer. In ARTSCAN the elective treatment volume for node positive OPC varied significantly between centres due to local treatment principles. All patients with OPC in complete response after primary treatment were eligible for the present case-control study. Cases were patients with RR during five years follow-up. Patients with no recurrence were eligible as controls. Four controls per case were matched according to T-and N-stage. Mean (D-mean) and median (D-50%) dose for the lymph node level (LNL) of RR in the cases and the corresponding LNL in the controls were analysed with conditional logistic regression. OS was estimated with the Kaplan-Meier method and evaluated by multivariate Cox regression analysis. Results: There was a dose-dependent risk reduction for D-50% in the interval that represented elective treatment (40-50 Gy) (OR = 0.18, p < 0.05) and a trend in the same dose interval for D-mean (OR = 0.19, p = 0.07). OS rates at five years were 0.39 (0.24-0.65) for cases and 0.70 (0.62-0.81) for controls (p < 0.001). The Kaplan-Meier and the Cox regression analysis for cases categorised by delivered dose showed an inverse relationship between dose and survival. The cases with RR in a LNL outside planning target volume (PTV) (D-mean < 40 Gy) had an OS rate comparable to that of all patients, and those with RR in a LNL in PTVelective (D-mean 40-60 Gy) or PTVtumour (D-mean > 60 Gy) did significantly worse (p < 0.05). The same inverse relationship was also shown for a small subset of patient with known HPV-status, defined by over expression of p16 (p < 0.05). Conclusions: There was a significant risk reduction for RR of elective treatment. However the OS for patients with RR outside target volumes was not affected, with similar results for patients with HPV-positive OPC. This could be an argument for a prospective randomized study on limited elective target volumes in OPC