目的:对钼靶(mammography,MG)、磁共振(magnetic resonance image,MRI)及核素显像(positron emission tomo-graphy/single photon emission computed tomography,PET/SPECT)诊断乳腺癌的准确性进行Meta分析。方法:按照循证医学诊断性试验的评价标准筛选1986-2006年PubMed、OVID、Medline数据库中有关MG、MRI及PET/SPECT诊断乳腺癌的文献资料,对筛选文献中3种诊断方法的相关数据和指标分别进行Meta分析,比较各自的诊断准确性。结果:共筛选出30篇文献,包含41组研究,其中14组关于MG(2 941个病灶),10组关于MRI(1 428个病灶),17组关于PET/SPECT(2 247个病灶)。Meta分析结果表明:MG、MRI、PET/SPECT诊断的敏感度分别为82%、86%和87%;特异度分别为69%、65%和82%;SROC曲线下面积(area under curve,AUC)分别为0.84、0.89和0.90;Q*值分别为0.77、0.81和0.88。3种检查方法的敏感度相似,无显著差异;PET/SPECT的特异度高于MG和MRI(P〈0.05);PET/SPECT和MRI的综合诊断效能大于MG。结论:MG仍然是目前乳腺癌初诊较合适的影像方法,年轻女性可优先考虑MRI检查,PET/SPECT可在其他检查疑似乳癌但不能明确诊断的情况下适当选用。
Objective: To compare the performance of mammography, magnetic resonance image and positron emission tomography/single photon emission computed tomography (PET/SPECT) in the diagnosis of breast cancer by Meta analysis. Methods: Based on the assessment standard of evidence-based diagnostic experiment, we searched PubMed, OVID, and Medline database (1986-2006) for the articles on the diagnosis of breast cancer with MG, MRI, PET and/or SPECT. The data and parameters of the 3 diagnostic methods were subjected to Meta analysis. Using software Excel 2003, SPSS13.0 and RevMan 4.2, we calculated the summary sensitivity and specificity, the pooled odds ratio (OR) and likelihood (LR). Summary receiver operating characteristic curve (SROC) was drawn to evaluate accuracy of the 3 methods. Results: We obtained 30 articles, which ineluded 41 data sets: 14 on MG (2 941 lesions), 10 on MRI (1 428 lesions), and 17 on PET/SPECT (2 247 lesions). The pooled estimates for sensitivity of MG, MRI and PET/SPECT were 82%(95%CI:76%-86%), 86%(95%CI:83%-88%), and 87%(95%CI:83%-90%), respectively; for specificity were 69%(95%CI:62%-75%), 65%(95%CI:62%-69%), and 82% (95%CI:76%-86%), respectively. Their weightedAUC (area under curve) were0.84, 0.89 and0.90, respeetively;Q values were 0.77, 0.81 and 0.88, respectively. The sensitivities of the 3 methods were not significantly different. The specificity of PET/SPECT was higher than those of MG and MRI (P〈0.05). The overall diagnostic effieaeies of PET/SPECT and MRI were better than that of MG. Conclusion: MG remains to be a satisfactory method for initial diagnosis of breast cancer; MRI examination should first be considered for younger patients. PET/SPECT should be used when a diagnosis of breast cancer can not be confirmed.