基于MRI及临床参数的鼻咽癌远处转移风险评估函数构建与评价

Construction and evaluation of MRI and clinical parameters-based risk assessment function for distant metastasis of nasopharyngeal carcinoma

目的探讨判别函数对初发鼻咽癌(NPC)患者远处转移风险评估的临床应用价值,为临床治疗方案制订提供参考依据。方法搜集具有完整影像检查及临床资料且随访12个月以上的初发NPC患者131例(111例无肿瘤远处转移,20例肿瘤远处转移);采用Fisher判别分析和Bayes判别分析对患者性别、年龄、是否涕血、T分期因素、N分期因素、T分期、N分期、临床分期、治疗前外周血生化指标及病理类型等变量进行逐步判别,筛选出具有显著判别能力的指标并建立判别函数,采用回顾性误判分析及交叉核实法误判分析对函数判别效能进行评价。结果颈部Ⅳ区淋巴结转移(X_(1))、肿瘤椎前肌侵犯(X_(2))、外周血小板计数(X_(3))和嗜酸性粒细胞绝对值(X_(4)),具有显著判别能力(P<0.05),并引入函数。Fisher判别分析得出1组函数为Y=1.116+1.588X_(1)+1.242X_(2)-0.010X_(3)+1.669X_(4),判别界值为0.4545。Bayes判别分析得出2组函数为Y_(1)=-8.10+0.277X_(1)+0.832X_(2)+0.058X_(3)-1.93X_(4)、Y_(2)=-7.234+2.354X_(1)+2.457X_(2)+0.044X_(3)+0.254X_(4)。回顾性误判分析结果显示,2组函数肿瘤远处转移和无远处转移的判别符合率分别为74.8%、75.0%,总体符合率为74.8%;交叉核实法误判分析结果显示,2组函数肿瘤远处转移和无远处转移的判别符合率分别为73.9%、70.0%,总体符合率为73.3%。结论判别函数可以作为NPC患者治疗前远处转移风险评估的有效补充手段,为临床治疗方案制订提供参考依据。

Objective To explore the clinical application value of discriminant function in assessing the risk of distant metastasis in patients with primary nasopharyngeal carcinoma(NPC)and to provide a reference basis for clinical treatment planning.Methods A total of 131 primary NPC patients(111 without and 20 with distant metastasis)with complete imaging and clinical data and more than twelve months of follow-up were collected.Fisher′s discriminant analysis and Bayes′discriminant analysis were used to stepwise discriminate between variables such as patients′gender,age,bleeding,T-stage factors,N-stage factors,T-stage,N-stage,clinical stage,peripheral blood biochemical indexes before treatment and pathological type.Indicators with significant discriminatory power were screened and discriminant functions were established.The discriminatory power of the function was evaluated by using retrospective misclassification analysis and cross-check misclassification analysis.Results Lymph node metastasis in cervical regionⅣ(X_(1)),tumor anterior vertebral muscle invasion(X_(2)),peripheral platelet count(X_(3))and absolute eosinophil value(X_(4))had significant discriminant power(P<0.05),and were introduced function.Fisher′s discriminant analysis yielded a set of functions:Y=1.116+1.588X_(1)+1.242X_(2)-0.01X_(3)+1.669X_(4),with a discriminant cut-off value of 0.4545.Bayes′discriminant analysis yielded two set of functions:Y_(1)=-8.10+0.277X_(1)+0.832X_(2)+0.058X_(3)-1.93X_(4),Y_(2)=-7.234+2.354X_(1)+2.457X_(2)+0.044X_(3)+0.254X_(4).The results of the retrospective misclassification analysis showed that the discrimination rates for the two set of functions were 74.8%and 75.0%for distant and no distant metastases respectively,with an overall compliance rate of 74.8%.The results of the cross-verification method misclassification analysis showed that the discriminatory compliance rates for distant and no distant metastases were 73.9%and 70.0%for the two sex of functions,respectively,and the overall compliance rate was 73.3%.Conclusion The discriminant function can be used as an effective complementary tool to assess the risk of distant metastasis in patients with NPC before treatment.It provides a reference basis for clinical treatment planning.