Risk stratification on the basis of Deauville score on PET-CT and the presence of Epstein-Barr virus DNA after completion of primary treatment for extranodal natural killer/T-cell lymphoma, nasal type: a multicentre, retrospective analysis

Background Assessment of tumour viability after treatment is essential for prediction of treatment failure in patients with extranodal natural killer/T-cell lymphoma (ENKTL). We aimed to assess the use of the post-treatment Deauville score on PET-CT and Epstein-Barr virus DNA as a predictor of residual tumour, to establish the risk of treatment failure in patients with newly diagnosed ENKTL. Methods In a retrospective analysis of patient data we assessed the prognostic relevance of the Deauville score (five-point scale) on PET-CT and circulating Epstein-Barr virus DNA after completion of treatment in consecutive patients with ENKTL who met eligibility criteria (newly diagnosed and received non-anthracycline-based chemotherapy, concurrent chemoradiotherapy, or both together) diagnosed at the Samsung Medical Center in Seoul, South Korea. The primary aim was to assess the association between progression-free survival and risk stratification based on post-treatment Deauville score and Epstein-Barr virus DNA. With an independent cohort from two different hospitals (Hong Kong and Singapore), we validated the prognostic value of our risk model. Findings We included 102 patients diagnosed with ENKTL between Jan 6, 2005, and Nov 18, 2013, in the study cohort, and 38 patients diagnosed with ENKTL between Jan 7, 2009, and June 27, 2013, in the validation cohort. In the study cohort after a median follow-up of 47·2 months (IQR 30·0–65·5), 45 (44%) patients had treatment failure and 33 (32%) had died. Post-treatment Deauville score and Epstein-Barr virus DNA positivity were independently associated with progression-free and overall survival in the multivariable analysis (for post-treatment Deauville score of 3–4, progression-free survival hazard ratio [HR] 3·607, 95% CI 1·772–7·341, univariable p<0·0001; for post-treatment Epstein-Barr virus DNA positivity, progression-free survival HR 3·595, 95% CI 1·598–8·089, univariable p<0·0001). We stratified patients into three groups based on risk of treatment failure: a low-risk group (post-treatment Epstein-Barr virus negativity and post-treatment Deauville score of 1–2), a high-risk group (post-treatment Epstein-Barr virus negativity with a Deauville score 3–4, or post-treatment Epstein-Barr virus positivity with a Deauville score 1–2), and treatment failure (Deauville score of 5 or post-treatment Epstein-Barr positivity with a Deauville of score 3–4). This risk model showed a significant association with progression-free survival (for low risk vs high risk, HR 7·761, 95% CI 2·592–23·233, p<0·0001; for low risk vs failure, HR 18·546, 95% CI 5·997–57·353, p<0·0001). The validation cohort showed the same associations (for low risk vs high risk, HR 22·909, 95% CI 2·850–184·162, p=0·003; for low risk vs failure, HR 50·652, 95% CI 6·114–419·610, p<0·0001). Interpretation Post-treatment Deauville score on PET-CT scan and the presence of Epstein-Barr virus DNA can predict the risk of treatment failure in patients with ENKTL. Our results might be able to help guide clinical practice. Funding Samsung Biomedical Research Institute.