Introduction

Breast cancer remains one of the leading causes of cancer-related deaths in women worldwide, emphasizing the importance of early detection for improving patient outcomes. Mammography has been the cornerstone of breast cancer screening for decades, but its sensitivity is limited, especially in women with dense breast tissue. As such, there has been increasing interest in alternative imaging modalities to enhance diagnostic accuracy.

Contrast-enhanced mammography (CEM) is a relatively new imaging technique that combines traditional mammography with intravenous contrast to improve the visualization of vascular characteristics within breast lesions. This method can provide insights into lesion perfusion and angiogenesis, characteristics that are often indicative of malignancy. CEM offers several potential advantages, including the ability to be performed using standard mammography equipment and the lower cost compared to MRI.

Magnetic resonance imaging (MRI), known for its superior soft tissue contrast, is considered a highly sensitive modality for detecting breast lesions, especially in women with dense breasts or those at high risk for breast cancer. MRI, however, is more expensive and less accessible, requiring specialized equipment and longer examination times.

This meta-analysis aims to systematically review the existing literature comparing CEM and MRI for breast lesion detection, focusing on sensitivity, specificity, and overall diagnostic performance. We hypothesize that MRI will demonstrate superior sensitivity, but CEM may provide comparable specificity, making both modalities important tools in the clinical setting.

Methods

1. Search Strategy and Inclusion Criteria

A systematic search was conducted using PubMed, Cochrane Library, and Scopus databases. The search terms included "contrast-enhanced mammography," "magnetic resonance imaging," "breast lesion," and "diagnostic accuracy." Studies were selected based on the following inclusion criteria:

• Studies comparing CEM and MRI for breast lesion detection.
• Studies that report diagnostic performance metrics, including sensitivity, specificity, PPV, NPV, and diagnostic accuracy.
• Studies published in peer-reviewed journals from 2000 to the present.
• Full-text availability in English.

Exclusion criteria included:

• Studies not involving human participants.
• Studies with insufficient data or unclear diagnostic outcomes.
• Case reports, letters, or abstracts.

2. Data Extraction and Quality Assessment

Two independent reviewers extracted the following data:

• Study Characteristics: First author, year of publication, study design, sample size, and lesion characteristics.
• Diagnostic Performance: Sensitivity, specificity, PPV, NPV, and diagnostic accuracy for both CEM and MRI.
• Lesion Type: Benign vs malignant.
• Imaging Protocol: Details on contrast agents, imaging parameters, and patient demographics.

The quality of each included study was assessed using the QUADAS-2 (Quality Assessment of Diagnostic Accuracy Studies) tool, which evaluates risk of bias and applicability concerns across four domains: patient selection, index test, reference standard, and flow and timing.

3. Statistical Analysis

The diagnostic performance of CEM and MRI was summarized by pooled sensitivity, specificity, PPV, NPV, and diagnostic accuracy. A random-effects model was used to account for between-study heterogeneity. The correlation coefficient (Pearson's r) was calculated to assess the relationship between CEM and MRI for detecting breast lesions. Heterogeneity was quantified using the I² statistic. An I² value of 0-25% indicated low heterogeneity, 25-75% moderate heterogeneity, and 75-100% high heterogeneity.

A summary receiver operating characteristic (ROC) curve was constructed to visualize and compare the diagnostic accuracy of CEM and MRI. Subgroup analyses were performed based on lesion type (benign vs malignant) and study design (prospective vs retrospective). Publication bias was assessed using funnel plots and Egger’s test.

Statistical analysis was conducted using RevMan 5.4 and R (version 4.0.3).