Logistic回归和ROC曲线评价癌胚抗原、鳞状细胞癌抗原和铁蛋白对肺癌的诊断价值

摘要

Objective To investigate the diagnostic value of serum CEA, SCC and SF as analyzed with Logistic regres-sion and ROC curve in patients with lung cancer. Methods Serum CEA, SCC and SF were measured in 100 patients with lung cancer (lung cancer group) and 30 healthy subjects (healthy group). Logistic regrssion and ROC curve were applied to analyze the data and evaluate the diagnostic values. Results The concentration of serum CEA, SCC and SF in lung cancer group were significantly higher than the healthy group [(13.82±21.42) μg/L v s (2.79±1.39) μg/L; (3.93±7.58)μg/L vs (0.76±0.28)μg/L;(455.31±271.38)μg/L vs (148.50±97.30)μg/L, P<0.01]. The serum level of CEA was the highest in patients with adenocarcinoma [(26.42±28.97) μg/L]. SCC was the highest in squamous cell lung cancer [(8.01±11.32)μg/L]. The concentrations of three tumor markers correlated strongly with tumor stage, the serum levels of markers in patients with stage Ⅲ and Ⅳ were significantly higher than in those with stageⅠand II [(15.17±23.20)μg/L v s (8.72±11.70)μg/L;(4.57±8.42)μg/L v s (1.53±0.86)μg/L;(479.08±280.72)μg/L v s (365.90±215.86)μg/L, P<0.05, P< 0.01], showing significantly progressively increasing concentrations in advancedcases. According to regrssion equation Y=1/[1+EXP(2.261 X1+3.459 X2+4.267 X3-1.082)], and the A UC of variable Y was higher than any one of the three tumor markers. Conclusion Combined detection CEA, SCC and SF can improve the diagnostic sensitivity and ac-curacy significantly. Applications of Logistic regression and ROC curve increase diagnostic accuracy in lung cancer.%目的：探讨Logistic回归和ROC曲线综合分析血清癌胚抗原(CEA)、鳞状细胞癌抗原(SCC)和铁蛋白(SF)对肺癌的诊断价值。方法采用电化学发光法检测CEA，酶联免疫吸附测定法检测SCC，免疫比浊法检测SF，检测100例肺癌患者(肺癌组)和30例健康者(健康组)血清中的CEA、SCC、SF水平，通过Logistic回归建立回归模型，用ROC曲线分析三种肿瘤标志物对肺癌的诊断的意义。结果肺癌组CEA、SCC、SF水平显著高于健康组[(13.82±21.42)μg/L比(2.79±1.39)μg/L；(3.93±7.58)μg/L比(0.76±0.28)μg/L；(455.31±271.38)μg/L比(148.50±97.30)μg/L，均P<0.01]，腺癌患者CEA水平[(26.42±28.97)μg/L]最高，鳞癌患者SCC水平[(8.01±11.32)μg/L]最高，肺癌患者三种肿瘤标志物水平与TNM分期密切相关，(Ⅲ+Ⅳ)期明显高于(Ⅰ+Ⅱ)期[(15.17±23.20)μg/L比(8.72±11.70)μg/L；(4.57±8.42)μg/L比(1.53±0.86)μg/L；(479.08±280.72)μg/L比(365.90±215.86)μg/L，P<0.05或P<0.01]，分期越晚，三种肿瘤标志物水平越高。建立回归模型Y=1/[1+EXP(2.261X1+3.459X2+4.267X3-1.082)]，新变量Y 的A UC高于三种单一肿瘤标志物的A UC。结论血清CEA、SCC和SF对肺癌的诊断具有较高的价值，联检可提高对肺癌的诊断率，综合运用Logistic回归和ROC曲线分析可提高肺癌诊断的准确性。