Meta-analysis of MR imaging in the diagnosis of breast lesions

CRD summary

The authors concluded that magnetic resonance imaging had high sensitivity and lower specificity in patients referred for biopsy of a breast lesion. The authors’ conclusions appeared to reflect the evidence, but potential for bias in the review and heterogeneity among studies mean they should be interpreted with caution.

Authors' objectives

To assess the diagnostic accuracy of contrast material-enhanced magnetic resonance imaging (MRI) in patients with breast lesions.

Searching

MEDLINE was searched between 1985 and March 2005 for English-language publications. Search terms were reported. PubMed, DARE and The Cochrane Library were searched to identify relevant studies from meta-analyses and reviews. References were cross checked.

Study selection

Studies of women aged over 19 years and suspected of having breast cancer (particularly small lesions detected at mammographic screening) who underwent material-enhanced MRI and biopsy were eligible for inclusion. Eligible studies had to include at least one patient with a non-palpable lesion and report numbers of false and true positives and negatives. MRIs were required to have a field strength of at least 1.5T and have acquired T1-weighted images before and after the contrast agent was administered. No complex imaging tools (such as magnetic resonance spectroscopy or diffusion-weighted imaging) were to have been used. Studies were required to include an appropriate reference-standard (such as histological analysis and mammography and clinical follow-up of more than two years) in more than 75% of the study population.

Included studies were conducted in Europe, Korea, Japan, USA, Canada and Egypt between 1994 and 2004. Mean patient age ranged between 46 and 58 years. The proportion of patients with cancer ranged between 23% and 84%. Prevalence of palpable lesions ranged from 2% to 86%. Prevalence of invasive ductal carcinoma ranged from 23% to 100%. Most studies used the same type and dose of contrast agent. Most studies used an appropriate reference standard.

It was unclear how many reviewers screened studies for inclusion.

Assessment of study quality

Study quality was assessed with QUADAS (quality assessment of diagnostic accuracy studies) and STARD (standards for reporting of diagnostic accuracy) criteria. Studies were excluded from the review based on study quality; no other details were provided.

The authors did not state how many reviewers performed the validity assessment.

Data extraction

Two authors independently extracted numbers of true positives and negatives and false positives and negatives into 2x2 tables to calculate sensitivity and specificity values and 95% confidence intervals (CIs) and diagnostic odds ratios. Where cells in the table contained zero, 0.5 was added to all cells.

Discrepancies were referred to a third reviewer and resolved through consensus.

Methods of synthesis

A summary receiver operating characteristic (sROC) curve was plotted and goodness of fit of the sROC curve was assessed using the unweighted R² statistic. Overall sensitivity and specificity and 95% confidence ellipse were calculated using bivariate analysis (random-effects model). Bivariate analysis assessed the effects of 16 covariates on overall sensitivity and specificity.

Results of the review

Forty-four studies (n=5,653, range 14 to 821) were included in the review. All were cohort studies and all were blind to histological findings. In 25 studies the cut-off was decided before knowledge of the results. Thirty-five studies used a 100% appropriate reference standard.

Overall sensitivity was 0.90 (95% CI 0.88 to 0.92). Overall specificity was 0.72 (95% CI 0.67 to 0.77). DOR was 23.5 (95% CI 16.8 to 32.9). There was evidence of statistical heterogeneity (R² 0.12, p=0.024). Sensitivity was not significantly affected by any of the covariates assessed using bivariate analysis. However, specificity varied according to cancer prevalence: specificity decreased with increasing cancer prevalence (low prevalence 0.81, moderate 0.71, high 0.61); and number of criteria used to differentiate benign from malignant lesions, with higher specificity when two malignancy criteria were used (0.81) compared to one criterion (0.74) or three criteria (0.67).

Authors' conclusions

Magnetic resonance imaging had high sensitivity and lower specificity in patients referred for biopsy of a breast lesion.

CRD commentary

The review question and inclusion criteria were clearly defined. The literature search included a number of databases. As the search was restricted to English publications, potentially relevant studies may have been missed. Study quality was reported to have been assessed, but results were only partially reported. It was unclear whether each stage of the review process was undertaken in duplicate, which meant that reviewer error and bias could not be ruled out. There was evidence of clinical heterogeneity among patients. Appropriate methods were used to investigate statistical heterogeneity. However, the authors stated that the results of the covariate analysis should be interpreted with caution due to the large amount of missing data for some studies.

Although the authors’ conclusions appeared to reflect the evidence, potential for bias in the review and heterogeneity among studies mean they should be interpreted with caution.

Implications of the review for practice and research

Practice: The authors did not state any implications for practice.

Research: The authors stated that further research was needed to determine the diagnostic performance of breast MRI in patients with non-palpable breast lesions and to assess the effects of covariates on diagnostic performance.

Funding

None stated.

Bibliographic details

Peters NH, Borel Rinkes IH, Zuithoff NP, Mali WP, Moons KG, Peeters PH. Meta-analysis of MR imaging in the diagnosis of breast lesions. Radiology 2008; 246(1): 116-124. [PubMed: 18024435]

Original Paper URL

http://radiology.rsna.org/content/246/1/116.abstract

Indexing Status

Subject indexing assigned by NLM

MeSH

Breast Neoplasms /diagnosis; Female; Humans; Magnetic Resonance Imaging; Sensitivity and Specificity

AccessionNumber

12008000097

Database entry date

09/02/2011

Record Status

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.