| BackgroundNasopharyngeal carcinoma(NPC)is highly sensitive to radiation.Radiotherapy is the mainstay treatment modality for this disease.As recommended by National Comprehensive Cancer Network(NCCN)and European Society for Medical Oncology(ESMO)guidelines,chemotherapy combined with radiotherapy is a crucial development for treating stage Ⅱ-Ⅳ NPC.With the application of intensity-modulated radiation therapy(IMRT),patients with NPC could achieve superb treatment outcome,with the 5-year overall survival(OS)reported being over 90% for stage Ⅱ patients.Some investigators even argued that chemotherapy could be omitted in stage Ⅱ patients.However,it might be arbitrary to remove chemotherapy for all stage Ⅱ patients.Some published studies found that significant heterogeneities in treatment outcomes were observed for stage Ⅱ patients and the risk of distant metastasis(DM)was relatively high in some particular patients,for whom it seems less suitable for deintensification strategies that omit chemotherapy.Thus,searching robust predictive factors of a high risk of DM for stage Ⅱ patients are urgently needed.This retrospective study was conducted to investigate the prognostic value of radiologic extranodal extension(rENE)in stage Ⅱ NPC.MethodsBetween June 2005 and December 2011,a total of 397 stage Ⅱ NPC patients with histologically diagnosed NPC were treated by IMRT.All patients were reclassification according to the 8th American Joint Committee on Cancer(AJCC)/Union for International Cancer Control(UICC)staging system.The acquired magnetic resonance imaging(MRI)were interpreted by Y.X.who was specialized in head and neck MRI with more than 10 years of experience.Stage Ⅱ NPC patients with N1 category(n =365)were enrolled and divided into three groups according to the situation of rENE:without rENE,suspected rENE,and confirmed rENE(grades: A,infiltration into surrounding fat;B,matted nodes;C,infiltration into adjacent structures). In order to detect statistically significant differences in inter-rater and intra-rater agreements,100 cases were randomly selected for inter-rater and intra-rater reliability assessments of rENE.The primary endpoint was distant metastasis-free survival(DMFS);additional endpoints were OS and regional relapse-free survival(RRFS).ResultsThe consistency index of without,suspected,grade A,grade B and grade C were0.700,0.293,0.513,0.861 and 1.0 for the inter-rater,and 0.820,0.621,0.733,0.919 and 1.0 for the intra-rater.There is no difference in the 7-year DMFS between suspected rENE and without rENE(93.5% vs.89.3%,P=0.332).Therefore,patients without rENE and suspected rENE were combined as one group(non-confirmed rENE).The DMFS of grade A was similar to non-confirmed rENE(94.7% vs.92.6%,p=0.650),and superior to patients with grade B and C(P <0.05).Only high-grade rENE(including matted nodes and infiltration into adjacent structures)could significantly influence the survival outcomes,patients with high-grade rENE had significantly poorer survival than those without,with the 7y-DMFS and OS demonstrated to be 78.5% vs.93.0%(P < 0.001)and 81.9% vs.89.9%(P = 0.050),respectively.ConclusionsHigh-grade rENE(including matted nodes and infiltration into adjacent structures),as defined in our study,is a stable criterion,with high intra-rater and inter-rater consistency.High-grade rENE was an evaluable predictor that could help with the selection of stage Ⅱ patients with high risk of distant metastasis.Background ENE has been introduced as a new parameter in the N classification in the AJCC/ UICC 8th edition TNM for non-viral related head and neck cancer(HNC).The identification of ENE for NPC was based on imaging but not pathology.Then can the rENE be incorporated into the N staging of the AJCC/UICC of NPC to improve the prognostication of distant metastasis(DM)risk? The prognostic value of rENE have been confirmed in the previous studies of stage Ⅱ NPC.Based on this finding,we conducted this retrospective study to evaluated the prognostic value of various grades of rENE and their potential roles in N-classification refinement for NPC.Methods All NPC patients treated with IMRT in our institution between June 2005 and December 2011 were included.Pre-treatment MRI of c N+ cases were reviewed and rENE was recorded as G0: lymph nodes without rENE;G1: tumor infiltrating beyond individual nodal capsule into the surrounding fat plane;G2: coalescent nodal mass with unequivocal evidence of rENE;G3: tumor infiltrating beyond nodal capsule into adjacent structures.COX multivariable analysis assessed prognostic value of rENE for DM and death adjusted for age,gender,lactate dehydrogenase(LDH),T-classification,N-classification,and chemotherapy cycles.We also evaluated that reliability(interrater concordance)and reproducibility(intra-rater concordance)of various grades of rENE by randomly selecte 100 cases.The prognostic performance of refined Nclassification by incorporating rENE grades was compared with TNM8-N using Akaike Information Criterio(AIC)and Concordance index(C-index)where lower AIC or high C-index indicated more robust models.The primary endpoint was DMFS;additional endpoints were OS and Locoregional Control(LRC).Results A total of 1390 of 1616(86%)NPC were c N+,and rENE was detected in 826/1390(59%)patients: 256(18.4%)G1-rENE,487(35%)G2-rENE,and 83(6%) G3-rENE.COX multivariable analysis confirmed that G2-/G3-rENE had increased risk of DM(hazard ratio(HR): 2.05/3.18,both P < 0.001)and death(HR: 1.62/2.39,P = 0.002/ P < 0.001),while G1-rENE was non-prognostic(DM: P = 0.172;death: P = 0.320).We propose a refined N: New＿N1: N1/N2 without G2-/ G3-rENE;New＿N2: N1＿G2-rENE;New＿N3: N2＿G2-rENE,N1/N2＿G3-rENE,or N3.The New＿N classification had a lower AIC and higher C-index for DM(AIC: 3809.6 vs.3830.9;C-index: 0.700 vs.0.677)and death(AIC: 3693.8 vs.3705.9;C-index: 0.735 vs.0.725)versus TNM-8 N.Conclusions G2-rENE and G3-rENE are independently prognostic for DM and death in NPC.Compared to the TNM8 N-classification,a refined N-classification incorporating G2-rENE and G3-rENE improves prognostication of DM and mortality risk. |
PDF Full Text Request