Plasma Epstein-Barr virus DNA and residual disease after radiotherapy for undifferentiated nasopharyngeal carcinoma.

Epstein-Barr virus (EBV) DNA can be detected and quantified in the plasma of patients with EBV-related tumors, such as nasopharyngeal carcinoma (NPC). Although NPC at early stages can be cured by radical radiotherapy, there is a high recurrence rate in patients with advanced NPC. The pretreatment level of circulating EBV DNA is a prognostic factor for NPC, but the prognostic value of post-treatment EBV DNA has not been studied. We designed a prospective study in Hong Kong, China, to investigate the value of plasma EBV DNA as a prognostic factor for NPC. One hundred seventy NPC patients, without metastatic disease at presentation, were treated with a uniform radiotherapy protocol. Circulating EBV DNA was measured by real-time quantitative polymerase chain reaction before treatment and 6-8 weeks after radiotherapy was completed. Risk ratios (RRs) were determined with a Cox regression model, and associations of various factors with progression-free and overall survival and recurrence rates were determined with a stepwise Cox proportional hazards model. All statistical tests were two-sided. Ninety-nine percent of patients achieved complete clinical remission. Levels of post-treatment EBV DNA dominated the effect of levels of pretreatment EBV DNA for progression-free survival. The RR for NPC recurrence was 11.9 (95% confidence interval [CI] = 5.53 to 25.43) for patients with higher post-treatment EBV DNA and 2.5 (95% CI = 1.14 to 5.70) for patients with higher pretreatment EBV DNA. Higher levels of post-treatment EBV DNA were statistically significantly associated with overall survival (P<.001; RR for NPC recurrence = 8.6, 95% CI = 3.69 to 19.97). The positive and negative predictive values for NPC recurrence for a higher level of post-treatment EBV DNA were 87% (95% CI = 58% to 98%) and 83% (95% CI = 76% to 89%), respectively. Levels of post-treatment plasma EBV DNA in patients with NPC appear to strongly predict progression-free and overall survival and to accurately reflect the post-treatment residual tumor load.