摘要Background:Non-Hodgkin T/NK cell lymphoma is a rare and widely variable type of lymphoma with the most dismal prognosis.This study aimed to investigate varied impact of the clinical indicators to the overall survival (OS).Methods:We conducted a retrospective study to identify the non-invasive clinical features of T cell lymphoma that can predict prognosis with an innovative analysis method using quantile regression.A total of 183 patients who visited a top-tier hospital in Beijing,China,were enrolled from January 2006 to December 2015.Demographic information and main clinical indicators were collected including age,erythrocyte sedimentation rate (ESR),survival status,and international prognostic index (IPI) score.Results:The median age of the patients at diagnosis was 45 years.Approximately 80％ of patients were at an advanced stage,and the median survival time after diagnosis was 5.1 months.Multivariable analysis of the prognostic factors for inferior OS associated with advanced clinical staging [HR=3.16,95 ％ CI (1.39-7.2)],lower platelet count [HR =2.57,95 ％ CI (1.57-4.19),P ＜ 0.001] and higher IPI score [HR=1.29,95％CI (1.01-1.66),P=0.043].Meanwhile,T cell lymphoblastic lymphoma [HR=0.40,95％CI (0.20-0.80),P=0.010],higher white blood cell counts [HR=0.57,95％CI (0.34-0.96),P=0.033],higher serum albumin level [HR=0.6,95％CI (0.37-0.97),P=0.039],and higher ESR [HR=0.53,95％CI (0.33-0.87),P=0.011] were protective factors for OS when stratified by hemophagocytic lymphohistiocytosis (HLH).Multivariable quantile regression between the OS rate and each predictor at quartiles 0.25,0.5,0.75,and 0.95 showed that the coefficients of serum β32-microglobulin level and serum ESR were statistically significant in the middle of the coefficient curve (quartile 0.25-0.75).The coefficient of IPI was negatively associated with OS.The coefficients of hematopoietic stem cell transplantation (HSCT) and no clinical symptoms were higher at the middle of the quartile level curve but were not statistically significant.Conclusions:The IPI score is a comparatively robust indicator of prognosis at 3 quartiles,and serum ESR is stable at the middle 2 quartiles section when adjusted for HLH.Quantile regression can be used to observe detailed impacts of the predictors on OS.