|Year : 2017 | Volume
| Issue : 5 | Page : 862-868
Meta-analysis of diagnostic accuracy of magnetic resonance imaging and mammography for breast cancer
Yueqiao Zhang, Hong Ren
Department of Radiology, School of Medicine, Sir Run Run Shaw Hospital, Zhejiang University, Hangzhou, China
|Date of Web Publication||13-Dec-2017|
Department of Radiology, School of Medicine, Sir Run Run Shaw Hospital, Zhejiang University, Hangzhou 310016
Source of Support: None, Conflict of Interest: None
Objective: The aim of this study was to compare the performance of mammography (MG) and magnetic resonance imaging (MRI) in the diagnosis of breast cancer.
Methods: Searching in the databases including PubMed, Embase, and Google Scholar about comparative study of MG and MRI in the diagnosis of breast cancer during 2000–2017. After we screened further, the extracted effective data were calculated by Meta-Disc 1.4 software.
Results: we obtained 11 articles. The pooled estimates for sensitivity of MG and MRI were 0.75 (95% confidence interval [CI], [0.72, 0.78]) and 0.92 (95% CI, [0.89, 0.94]) respectively, and for specificity were 0.71 (95% CI, [0.67, 0.74]) and 0.70 (95% CI, [0.66, 0.73]), respectively. Their weighted area under the summary receiver operating characteristic curve was 0.79 and 0.93, respectively.
Conclusion: MRI remains to be a satisfactory method for the diagnosis of breast cancer and should first be considered for patients.
Keywords: Breast cancer, mammography, magnetic resonance imaging, meta-analysis
|How to cite this article:|
Zhang Y, Ren H. Meta-analysis of diagnostic accuracy of magnetic resonance imaging and mammography for breast cancer. J Can Res Ther 2017;13:862-8
| > Introduction|| |
Breast cancer is the most common cancer and the fifth most common cause of death among women in the world. Improving the diagnosis way to detect the progression of breast cancer more early and accurate, this will be contributed to treatment of breast cancer patients. One of the key problems in the diagnosis of breast cancer is how distinguish breast cancer with some benign lesions such as breast fibroma, fibroadenosis, and lobular hyperplasia. The mammography (MG) is a traditional imaging tool in breast disease detection. In recent years, magnetic resonance imaging (MRI) has also been applied to the differential diagnosis and clinical research of breast cancer, which not only enriches the diagnosis method of breast cancer but also brings a series of question such as rational use of these methods, how to optimize their combination to achieve the best diagnostic results. Therefore, it is necessary to systematical evaluate the diagnostic accuracy and characteristics of the above two examination methods. This will provide valuable guidance for the selection of breast imaging methods and the interpretation of test results.
| > Methods|| |
We conducted a search of PubMed, Embase, and Google Scholar databases that were published between 2000 and 2017. We limited the search to study published in English. The medical subject heading terms and keywords used included “breast cancer or breast carcinoma or breast neoplasm,” “mammography,” “magnetic resonance image,” “sensitivity,” “specificity,” “accuracy,” and “diagnostic value.” Duplicate articles and unpublished studies from international meetings were excluded.
Studies were selected carefully on the basis of following criteria: The pathology results as “gold standard;” a case–control study (there are 2 × 2 contingency tables); blind evaluation results; direct or indirect access to true positive, false positive, false negative, true negative, sensitivity, specificity.
Studies were excluded on the basis of following criteria: nonclinical controlled trials; nonbreast cancer research; incomplete data; case reports; review literature; data published repeatedly.
Two authors independently assessed each literature and then download and extracted all the data using standardized data-abstraction forms. The data extracted included year of publication, true positive, false positive, false negative, true negative, sensitivity, and specificity. For each study, 2 × 2 contingency tables were constructed. We calculated the sensitivity, specificity, and likelihood ratio (LR).
The sensitivity, specificity, and 95% confidence interval (CI) of the literature were analyzed by Meta-Disc 1.4 software (Clinical BioStatistics Unit-Hospital Ramon y Cajal, Madrid, Spain), and the summary receiver operating characteristic (SROC) curves and forest map were drawn. According to the results of heterogeneity test, the corresponding fixed effect model or the immediate effect model is selected to quantify the effect values.
| > Results|| |
Literature searches and characteristics of eligible study
According to search strategy, 11 full articles were finally considered eligible for the review after evaluation. [Figure 1] shows the flow diagram of study selection process. The detailed characteristics for the 11 eligible studies are summarized in [Table 1].
We assessed the quality of included studies according to QUADAS. Each study was evaluated respectively by two independent investigators. On average, the investigators disagreed on three of 11 items (range, 0–6). All disagreements were resolved by consensus.
The diagnostic sensitivity and specificity of mammography
The pooled diagnostic sensitivity and specificity of MG are 0.75 (95% CI, [0.72, 0.78]) and 0.71 (95% CI, [0.67, 0.74]), respectively. Significant heterogeneity was found among these studies (I2 = 79.4% and 84.1%). Due to significant heterogeneity of the data, we used a random effects model [Figure 2] and [Figure 3].
The diagnostic sensitivity and specificity of magnetic resonance imaging
The pooled diagnostic sensitivity and specificity of MRI are 0.92 (95% CI, [0.89, 0.94]) and 0.70 (95% CI, [0.66, 0.73]), respectively. Significant heterogeneity was found among these studies (I2 = 84.8% and 94.1%). Due to significant heterogeneity of the data, we used a random effects model [Figure 4] and [Figure 5].
The negative likelihood ratio, positive likelihood ratio, and summary receiver operating characteristic curve of mammography
The pooled negative LR and positive LR of MG are 0.39 (95% CI, [0.29, 0.52]) and 0.71 (95% CI, [0.67, 0.74]), respectively [Figure 6] and [Figure 7]. We successfully plotted the SROC curve. The area under the SROC curve (AUC) is 0.79 and the Q* is 0.73 [Figure 8].
|Figure 8: The summary receiver operating characteristic curve of mammography|
Click here to view
The negative likelihood ratio, positive likelihood ratio, and summary receiver operating characteristic curve of magnetic resonance imaging
The pooled negative LR and positive LR of MG are 0.13 (95% CI, [0.06, 0.29]) and 3.64 (95% CI, [2.29, 5.77]), respectively [Figure 9] and [Figure 10]. We successfully plotted the SROC curve. The AUC is 0.93 and the Q* is 0.87 [Figure 11].
|Figure 9: The plot for the negative likelihood ratio of magnetic resonance imaging|
Click here to view
|Figure 10: The plot for the positive likelihood ratio of magnetic resonance imaging|
Click here to view
|Figure 11: The summary receiver operating characteristic curve of magnetic resonance imaging|
Click here to view
| > Discussion|| |
When the traditional literature review deals with the same problem of multiple studies, it is usually equal to treat each study results so that sample size and quality of each study are ignored. These are the major causes of heterogeneity. Therefore, it is difficult to ensure authenticity and reliability of the results. Meta-analysis is a systematic study of multiple-independent research results of the same research purpose, and it is a research method of quantitative synthesis. We gained multiple-independent research about investigating diagnostic accuracy of MRI and MG for breast cancer by searching databases. Meta-analysis can increase sample content and test effectiveness so that optimize clinical decision based on comprehensive information rather than personal experience. With the development of summary ROC method, the meta-analysis of the diagnostic test tends to improve.
Diagnostic tests usually used sensitivity and specificity as indicators. In the same experiment, the two indicators are inversely proportional each other. One indicator rises and the other indicator will decrease. Therefore, a single indicator is not able to reflect the whole character of the diagnostic test. ROC curve can reflect the characteristics of diagnostic tests. Meta-analysis of several different tests of the same test index can be expressed by an ROC curve according to the weight of their odds ratio. This curve is called the SROC curve. The AUC and Q* can be obtained by calculating the area under SROC curve. Q* is the point on the SROC where sensitivity and specificity are equal. The greater AUC and Q* show better diagnostic authenticity. In the sensitivity and specificity, we found sensitivity of 0.75 and specificity of 0.71 for breast MG. The sensitivity and specificity of breast MRI were 0.92 and 0.70. This showed that the discriminatory power of breast MRI is better than MG.
Diagnostic tests are often difficult to suit the principle of randomized controlled trials. It is almost impossible to randomize the “test” and “control” groups in the course of practice, and because different researchers often use different thresholds when judging the test results, most of the diagnostic results are heterogeneous. In this study, both MG group and MRI group showed heterogeneity, so the random effects model was used. This is also a accepted practice, which appropriately increase the weight of small sample data and reduce the weight of large sample data to deal with the heterogeneity of data, but this will bring a certain risk because the quality of small sample data is usually poor and greater biased, and the quality of large sample data is often better, less biased. The random effects model may affect the authenticity of results. On the basis of this meta-analysis, some diagnostic strategies can be used to guide the clinical: (1) although MG shows lower sensitivity than MRI, the cost of MG is cheaper than MRI. Hence, we believe that MG is still the first choice of breast cancer imaging method in clinical applications. (2) For young women, taking into account, the ionizing radiation on the breast damage and young women with dense breast ratio of the high physiological characteristics can give priority to MRI examination. (3) The LR is a combination factor of sensitivity and specificity that can more fully reflect the accuracy of diagnostic tests. The epidemiological studies suggest that LR+ >10 has a positive value and LR−<0.1 has a negative value for breast cancer diagnosis. The LR of two kinds of imaging methods is not very satisfactory. On the whole, any of them to diagnose ability of breast cancer is not enough. We still need to combine with other clinical data to analysis.
Evidence-based diagnosis is the use of “the best” research evidence for the diagnosis and treatment of patients to make the best decision. The evidence applied to clinical practice to solve practical problems. For MG, when the characteristics of certain diseases appear, such as “refined salt calcification” exhibited breast cancer, its sensitivity and specificity than the other methods. For dense breast, although MG has a difficulty to show the lesion, MRI should be considered to provide a soft tissue-rich level anatomical images. Although this study performed extensively to retrieve but failed to obtain unpublished literature, thus it could not rule out potential publication bias. In addition, further subgroup analysis was not carried out because that was limited by the initial information of included studies. The quality of original literature might also affect the meta-analysis results. At present, the diagnosis of breast lesions uses a variety of integrated imaging methods to improve the diagnostic accuracy. Their diagnostic value is not the same. Thorough further subgroup analysis may provide information that is instructive for diagnosis and identification.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| > References|| |
Zuo TT, Zheng RS, Zeng HM, Zhang SW, Chen WQ. Female breast cancer incidence and mortality in China, 2013. Thorac Cancer 2017;8:214-8.
Scenario IT. What is the role of nanotechnology in diagnosis and treatment of metastatic breast cancer? Promising scenarios for the near future. J Nanomaterials 2016;2016(5).
Guo C, Li X, Ye M, Xu F, Yu J, Xie C, et al.
Discriminating patients with early-stage breast cancer from benign lesions by detection of oxidative DNA damage biomarker in urine. Oncotarget 2017;8:53100-9.
Lalji U, Lobbes M. Contrast-enhanced dual-energy mammography: A promising new imaging tool in breast cancer detection. Womens Health (Lond) 2014;10:289-98.
Zhang J, Wang PJ, Yuan XD, Tian JM, Dong NX, Xu WG, et al
. Diagnosis accuracy of mammography, magnetic resonance image, and radionuclide imaging for breast cancer: A meta-analysis. Acad J Second Mil Med Univ 2007;28:1098-103.
Ng AK, Garber JE, Diller LR, Birdwell RL, Feng Y, Neuberg DS, et al.
Prospective study of the efficacy of breast magnetic resonance imaging and mammographic screening in survivors of Hodgkin lymphoma. J Clin Oncol 2013;31:2282-8.
Yabuuchi H, Matsuo Y, Sunami S, Kamitani T, Kawanami S, Setoguchi T, et al.
Detection of non-palpable breast cancer in asymptomatic women by using unenhanced diffusion-weighted and T2-weighted MR imaging: Comparison with mammography and dynamic contrast-enhanced MR imaging. Eur Radiol 2011;21:11-7.
Shao H, Li B, Zhang X, Xiong Z, Liu Y, Tang G, et al.
Comparison of the diagnostic efficiency for breast cancer in chinese women using mammography, ultrasound, MRI, and different combinations of these imaging modalities. J Xray Sci Technol 2013;21:283-92.
Ye F, Zeng M, Yan F, Feng W, Zhou M, Li R, Chen W. Clinical study of digital mammography, contrast-enhanced MRI as well as their combination in the diagnosis of breast cancer. Chin Ger J Clin Oncol 2008;7:286-91.
Malur S, Wurdinger S, Moritz A, Michels W, Schneider A. Comparison of written reports of mammography, sonography and magnetic resonance mammography for preoperative evaluation of breast lesions, with special emphasis on magnetic resonance mammography. Breast Cancer Res 2001;3:55-60.
Berg WA, Gutierrez L, NessAiver MS, Carter WB, Bhargavan M, Lewis RS, et al.
Diagnostic accuracy of mammography, clinical examination, US, and MR imaging in preoperative assessment of breast cancer. Radiology 2004;233:830-49.
Ozülker T, Ozülker F, Ozpaçaci T, Bender O, Deǧirmenci H. The efficacy of (99m) Tc-MIBI scintimammography in the evaluation of breast lesions and axillary involvement: A comparison with X-rays mammography, ultrasonography and magnetic resonance imaging. Hell J Nucl Med 2010;13:144-9.
Pediconi F, Catalano C, Roselli A, Dominelli V, Cagioli S, Karatasiou A, et al.
The challenge of imaging dense breast parenchyma: Is magnetic resonance mammography the technique of choice? A comparative study with x-ray mammography and whole-breast ultrasound. Invest Radiol 2009;44:412-21.
Vassiou K, Kanavou T, Vlychou M, Poultsidi A, Athanasiou E, Arvanitis DL, et al.
Characterization of breast lesions with CE-MR multimodal morphological and kinetic analysis: Comparison with conventional mammography and high-resolution ultrasound. Eur J Radiol 2009;70:69-76.
Sim LS, Hendriks JH, Fook-Chong SM. Breast ultrasound in women with familial risk of breast cancer. Ann Acad Med Singapore 2004;33:600-6.
Boné B, Wiberg MK, Szabó BK, Szakos A, Danielsson R. Comparison of 99mTc-sestamibi scintimammography and dynamic MR imaging as adjuncts to mammography in the diagnosis of breast cancer. Acta Radiol 2003;44:28-34.
Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The development of QUADAS: A tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol 2003;3:25.
Moses LE, Shapiro D, Littenberg B. Combining independent studies of a diagnostic test into a summary ROC curve: Data-analytic approaches and some additional considerations. Stat Med 1993;12:1293-316.
Walter SD. Properties of the summary receiver operating characteristic (SROC) curve for diagnostic test data. Stat Med 2002;21:1237-56.
Jackson D, White IR, Thompson SG. Extending derSimonian and laird's methodology to perform multivariate random effects meta-analyses. Stat Med 2010;29:1282-97.
Protassov RS, Van Dyk DA, editors. Detecting Multiple Model Components with the Likelihood Ratio Test. Bulletin of the American Astronomical Society. Astrophysical Journal 2000;32:761.
Sultana A, Ciuc M, Florea L, C Florea C, editors. Detection of mammographic microcalcifications using a refined region-growing approach. International Symposium on Signals, Circuits and Systems. 2009 International Symposium on Signals, Circuits and Systems; 2009. p. 1-4.
Upadhyay N, Saifuddin A. The radiographic and MRI features of gout referred as suspected soft tissue sarcoma: A review of the literature and findings from 27 cases. Skeletal Radiol 2015;44:467-76.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6], [Figure 7], [Figure 8], [Figure 9], [Figure 10], [Figure 11]