Objective: To assess the accuracy of computed tomographic colonography (CTC) for detection of colorectal neoplasm. Methods: Publications prior to January 2016 from the Medline, Embase, CNKI, WANFANG, and VIP literature databases were systematically reviewed. A QUADAS checklist was used to assess the quality of the studies. According to the sizes of tumor (≥6 mm or ≥10 mm), diagnostic test accuracy indexes (area under the curve, sensitivity and specific) were pooled and stratified. Spearman correlation and curve of summary receiver operating characteristic (SROC) were applied to comprehensively assess the threshold effect. A bi-variate mixed-effects model was used for testing the overall merging value and heterogeneity. Results: A total of 19 articles (n=11 540) were included in the analysis. A total of 18 studies were from Europe and Unite States, with 1 from Asia. Numbers of studies regarding information on general population, high-risk groups and clinical symptoms were 9, 5 and 5, respectively. In 19 articles, data on the accuracy of CTC diagnosis (≥6 mm group) was recorded, with another 17 on the group ≥10 mm. Area under the SROC curve in the ≥6 mm group was 0.92 (95% CI: 0.92-0.94). It was estimated that the pooled sensitivity and specificity were 0.80 (95% CI: 0.73-0.86) and 0.89 (95% CI: 0.86-0.92) respectively. In area under the SROC curve, the pooled sensitivity and specificity in ≥10 mm group were 0.87(95%CI:0.80-0.92), 0.97(95%CI:0.95-0.98), respectively. Results from the subgroup showed that the sensitivity of with or without the fecal tagging groups were 0.84 and 0.67 in the ≥6 mm group, 0.92 and 0.76 in the ≥10 mm group. It was also found that in the experienced or less experienced readers, rates of sensitivity were 0.83 and 0.75 in the ≥6 mm group, 0.91 and 0.79 in the ≥10 mm group. Conclusions: CTC had high diagnostic efficiency for colorectal neoplasm, especially in the ≥10 mm group. Fecal tagging and experience of the reasers can improve the diagnostic sensitivity.
目的： 对全球CT结肠成像（CTC）技术诊断结直肠肿瘤的准确度进行Meta分析。 方法： 在Medline、Embase、中国知网、万方数据知识服务平台和维普期刊资源整合服务平台检索1994年1月至2016年1月发表的相关文献，以QUADAS工具评价文献质量。按肿瘤大小分组（≥6 mm和≥10 mm），对诊断试验准确性指标（曲线下面积、灵敏度和特异度）进行汇总和分层分析；采用Spearman相关和综合受试者工作特征（SROC）曲线评估阈值效应，采用双变量混合效应模型进行效用值合并和异质性检验。 结果： 最终纳入19篇文献（11 540例），其中18篇来自欧洲和美国，1篇来自亚洲；9篇研究对象为普通人群，5篇高危人群，5篇有临床症状人群；19篇文献均报告CTC诊断≥6 mm结直肠肿瘤准确性数据，17篇报告诊断≥10 mm结直肠肿瘤准确性数据。Meta分析结果显示：肿瘤病变≥6 mm组，CTC用于结直肠肿瘤筛查的SROC曲线下的面积为0.92（95%CI：0.92~0.94），总体灵敏度为0.80（95%CI：0.73~0.86），总体特异度为0.89（95%CI：0.86~0.92）。亚组分析结果显示：在≥6 mm和≥10 mm组中，各因素对特异度影响均不明显，检查前是否进行粪便标记（≥6 mm组：是=0.84、否=0.67，≥10 mm组：是=0.92和否=0.76）、阅片人经验（≥6 mm组：丰富=0.83、不足=0.75，≥10 mm组：丰富=0.91、不足=0.79）对灵敏度有较大影响。 结论： CTC诊断结直肠肿瘤的效能较高，尤其是对≥10 mm肿瘤；进行粪便标记、丰富阅片人经验可提高诊断灵敏度。.
Keywords: Colorectal neoplasm; Computed tomographic colonography; Meta-analysis; Sensitivity and specificity.