Original Article
Establishing M1 stage subdivisions by incorporating radiological features and Epstein-Barr virus DNA for metastatic nasopharyngeal carcinoma
Abstract
Background: Significance of plasma Epstein-Barr virus deoxyribonucleic acid (EBV DNA)— a proven robust indicator for nasopharyngeal carcinoma (NPC)—is not yet clarified in risk stratification of metastatic NPC (mNPC). We aim to establish effective M1 stage subdivisions in mNPC by integrating radiological features and EBV DNA at diagnosis of metastasis (mEBV DNA).
Methods: The study comprised 1,007 mNPC patients, including 817 metachronous mNPC (mmNPC) patients randomized into training (n=613) and internal validation (n=204) cohorts, and 190 synchronous mNPC (smNPC) patients defined as smNPC validation cohort. Primary clinical end-point was overall survival (OS). Covariate inclusion to recursive partitioning analysis (RPA)-generated risk stratification was qualified by a multivariable two-sided P<0.05. Performances of different models were compared using area under ROC curve (AUC), Harrell’s concordance index (c-index) and Akaike information criterion (AIC).
Results: Compared with other simply image-based models, the ultimate RPA-EBV-stage presented a best performance [c-index =0.68 (training), 0.70 (internal validation), 0.64 (smNPC validation); AUC =0.69 (training), 0.72 (internal validation), 0.70 (smNPC validation)]: M1a (low mEBV DNA + oligo lesion), M1b (low mEBV DNA + multiple lesions), M1c (high mEBV DNA + no liver involvement), and M1d (high mEBV DNA + liver involvement). Corresponding 3-year OS rates were 49.9%, 33.4%, 22.6%, and 6.7%, respectively (P<0.001). In mmNPC patients, compared with chemotherapy alone, addition of local treatment demonstrated superiority in M1a and M1b; systemic therapy combined with targeted therapy conferred benefit on patients of M1c and M1d (P<0.05).
Conclusions: This RPA-EBV-stage provided favorable prognostic value for survival outcomes and could assist clinical and investigative management. Low-risk patients are considered suitable candidate for curative local treatment, and high-risk patients are recommended to undergo intensive systemic treatment.
Methods: The study comprised 1,007 mNPC patients, including 817 metachronous mNPC (mmNPC) patients randomized into training (n=613) and internal validation (n=204) cohorts, and 190 synchronous mNPC (smNPC) patients defined as smNPC validation cohort. Primary clinical end-point was overall survival (OS). Covariate inclusion to recursive partitioning analysis (RPA)-generated risk stratification was qualified by a multivariable two-sided P<0.05. Performances of different models were compared using area under ROC curve (AUC), Harrell’s concordance index (c-index) and Akaike information criterion (AIC).
Results: Compared with other simply image-based models, the ultimate RPA-EBV-stage presented a best performance [c-index =0.68 (training), 0.70 (internal validation), 0.64 (smNPC validation); AUC =0.69 (training), 0.72 (internal validation), 0.70 (smNPC validation)]: M1a (low mEBV DNA + oligo lesion), M1b (low mEBV DNA + multiple lesions), M1c (high mEBV DNA + no liver involvement), and M1d (high mEBV DNA + liver involvement). Corresponding 3-year OS rates were 49.9%, 33.4%, 22.6%, and 6.7%, respectively (P<0.001). In mmNPC patients, compared with chemotherapy alone, addition of local treatment demonstrated superiority in M1a and M1b; systemic therapy combined with targeted therapy conferred benefit on patients of M1c and M1d (P<0.05).
Conclusions: This RPA-EBV-stage provided favorable prognostic value for survival outcomes and could assist clinical and investigative management. Low-risk patients are considered suitable candidate for curative local treatment, and high-risk patients are recommended to undergo intensive systemic treatment.
