Abstract
BackgroundAlthough early-stage natural killer/T-cell lymphoma (NKTCL) typically responds to radiation therapy (RT) and asparaginase-based chemotherapy, treatment-related clonal evolution may eventually lead to therapeutic failure. Prognostic stratification based on baseline features fails to capture emerging resistance during therapy. Circulating tumor DNA (ctDNA) offers a potential tool for real-time monitoring, but its clinical utility in NKTCL remains unclear.MethodClinical data from 1,825 consecutive patients diagnosed with NKTCL between January 2014 and December 2022 were collected from five hospitals within the Chinese Southwest Oncology Group (CSWOG). Patients with early-stage NKTCL who received asparaginase-based induction chemotherapy (IC) followed by RT, along with longitudinal monitoring of cell-free EBV DNA (cfEBV-DNA), were included in the analysis. We assessed patterns of appearance, disappearance and re-appearance cfEBV-DNA which we correlated with progression-free survival (PFS) and overall survival (OS). The latter were assessed using the Kaplan-Meier method and compared using the log-rank test. Multi-variable Cox regression analysis was used to identify prognostic co-variates and hierarchical clustering analysis used to determine distinct cfEBV-DNA response patterns.ResultsA total of 710 patients were included. Pretreatment cfEBV-DNA was positive in 487 patients (68.6%), 258 of whom became negative after the first cycles of IC. During chemotherapy the percentage of patients becoming cfEBV-DNA-negative increased, whereas the velocity of cfEBV-DNA clearance decreased. Achieving cfEBV-DNA negative during different treatment phases (post-IC1 to post-IC6, and post-RT) was significantly associated with better PFS and OS compared with those who remained positive (all P < 0.01). Next, using unsupervised clustering analysis, patients were classified into four cfEBV-DNA response phenotypes: (1) consistently negative; (2) early response; (3) late response; and (4) no response. These cohorts had significantly different 5-year PFS (94% [95% Confidence Interval, 90, 98%], 83% [77, 88%], 57% [47, 68%] and 18% [11, 29%]; P-value for trend < 0.001) and OS (98% [96, 99%], 91% [87, 95%], 70% [60, 81%], and 33% [23, 45%]; P-value for trend < 0.001). In Cox multi-variable analyses cfEBV-DNA response phenotype was still independently correlated with PFS and OS. ConclusionIn conclusion, in patient with early-stage NKTCL, dynamic monitoring of cfEBV-DNA response and pattern thereof correlates with PFS and OS. Our study highlights the unique patterns of longitudinal on-treatment ctDNA and underscores its potential for real-time risk monitoring, providing valuable guidance for risk-based treatment strategies in NKTCL patients.
