论文部分内容阅读
目的评价EIA法检测血清神经元特异性烯醇酶诊断小细胞肺癌的价值。方法检索MEDLINE、EMbase、Cochrane图书馆等数据库,检索时间为1966年～2008年3月,收集关于EIA法检测血清神经元特异性烯醇酶诊断小细胞肺癌价值的相关研究文献并进行质量评价,采用Cochrane协作网提供的RevMan4.2.7软件进行异质性分析,MetaDisc进行Meta分析及绘制SROC曲线。结果共纳入15篇文献,包括4221例患者,其中经金标准确认的SCLC患者共672例,非SCLC者3549例。Meta分析结果显示,各纳入研究均存在不同程度的质量问题,导致各研究间存在异质性(P=0.0002,I2=66.1%),15个研究的合并灵敏度是0.67,95%CI(0.64,0.71),合并特异度是0.91,95%CI(0.90,0.92)。亚组分析结果显示,4个使用北京军事医学科学院放射医学研究所提供试剂的研究间未见统计学异质性(P=0.33,I2=13.4%),SROC曲线下面积0.9672,SE=0.0393;4个使用瑞士Roche公司提供试剂的研究间未见统计学异质性(P=0.23,I2=29.9%),SROC曲线下面积0.8311,SE=0.0836。结果表明其平均漏诊率为35.1%,平均误诊率为7.4%,有一定的诊断价值。结论现有有限证据表明:血清神经元特异性烯醇酶(NSE)在早期诊断小细胞肺癌中有一定的参考价值,可作为较重要的参考指标之一,但需要更多高质量的研究加以验证。
Objective To evaluate the value of EIA in the detection of serum neuron-specific enolase in the diagnosis of small cell lung cancer. Methods The databases of MEDLINE, EMbase and Cochrane Library were retrieved from 1966 to March 2008, and relevant research literature on the value of EIA in detecting small cell lung cancer by serum neuron-specific enolase was collected and evaluated. Heterogeneity analysis was performed using RevMan 4.2.7 software provided by the Cochrane Collaboration, Meta analysis by MetaDisc, and SROC curve mapping. Results A total of 15 articles were included, including 4221 patients, including 672 SCLC patients confirmed by gold standard and 3549 non-SCLC patients. The results of Meta-analysis showed that there were some quality problems in each study, leading to heterogeneity between the studies (P = 0.0002, I2 = 66.1%). The combined sensitivity of the 15 studies was 0.67 and 95% CI (0.64, 0.71) and the combined specificity was 0.91, 95% CI (0.90, 0.92). The results of subgroup analysis showed that there was no statistical heterogeneity (P = 0.33, I2 = 13.4%) in the four studies using reagents provided by the Institute of Radiation Medicine of Beijing Academy of Military Medical Sciences. The area under the SROC curve was 0.9672 and SE was 0.0393. There was no statistical heterogeneity among the four studies (R = 0.23, I2 = 29.9%) using Roche supplied reagents. The area under the SROC curve was 0.8311 and SE was 0.0836. The results showed that the average missed diagnosis rate was 35.1%, the average misdiagnosis rate was 7.4%, a certain diagnostic value. CONCLUSIONS There is limited evidence available to demonstrate that serum neuron-specific enolase (NSE) may be a valuable reference for the early diagnosis of small cell lung cancer and may require more high-quality studies verification.