Diagnostic Performance of Diffusion Tensor Imaging for Characterizing Breast Tumors: A Comprehensive Meta-Analysis.

Kai Wang,Yucong Zheng,Zhipeng Li,Zhifeng Wu,Linning E,Jianye Liang,Sihui Zeng

doi:10.3389/fonc.2019.01229

Abstract

Rationale and Objectives: Controversy still exists on the diagnosability of diffusion tensor imaging (DTI) for breast lesions characterization across published studies. The clinical guideline of DTI used in the breast has not been established. This meta-analysis aims to pool relevant evidences and evaluate the diagnostic performance of DTI in the differential diagnosis of malignant and benign breast lesions.Materials and Methods: The studies that assessed the diagnostic performance of DTI parameters in the breast were searched in Embase, PubMed, and Cochrane Library between January 2010 and September 2019. Standardized mean differences and 95% confidence intervals of fractional anisotropy (FA), mean diffusivity (MD), and three diffusion eigenvalues (λ1, λ2, and λ3) were calculated using Review Manager 5.2. The pooled sensitivity, specificity, and area under the curve (AUC) were calculated with a bivariate model. Publication bias and heterogeneity between studies were also assessed using Stata 12.0.Results: Sixteen eligible studies incorporating 1,636 patients were included. The standardized mean differences indicated that breast cancers had a significantly higher FA but lower MD, λ1, λ2, and λ3 than those of benign lesions (all P < 0.05). Subgroup analysis indicated that invasive breast carcinoma (IBC) had a significantly lower MD value than that of ductal carcinoma in situ (DCIS) (P = 0.02). λ1 showed the best diagnostic accuracy with pooled sensitivity, specificity, and AUC of 93%, 92%, and 0.97, followed by MD (AUC = 0.92, sensitivity = 87%, specificity = 83%) and FA (AUC = 0.76, sensitivity = 70%, specificity = 70%) in the differential diagnosis of breast lesions.Conclusion: DTI with multiple quantitative parameters was adequate to differentiate breast cancers from benign lesions based on their biological characteristics. MD can further distinguish IBC from DCIS. The parameters, especially λ1 and MD, should attract our attention in clinical practice.

Highlights

Breast cancer is the most commonly diagnosed cancer and the leading cause of cancer death among females in the world based on the GLOBOCAN 2018 estimates of cancer incidence and mortality [1]
Several studies indicated that diffusion weighted imaging (DWI)-derived apparent diffusion coefficient (ADC) values, named mean diffusivity (MD) in diffusion tensor imaging (DTI) protocol, significantly decreased in breast cancers compared with benign lesions, and it increased the ability of dynamic contrastenhanced (DCE)-MRI to differentiate cancers from benign lesions [4, 6]
The inclusion criteria were as follows: (a) DTI was used to differentiate breast cancer from benign lesions; (b) sufficient data regarding mean and standard deviation (SD) or diagnostic performance of DTI parameters [i.e., sensitivity, specificity, truepositive (TP), false-negative (FN), false-positive (FP), and truenegative (TN)] were reported or can be calculated from the study; (c) the breast lesions were confirmed by pathology; (d) the patients have not been treated with surgery or chemotherapy before magnetic resonance scanning; and (e) the scores of quality assessment based on likelihood of bias were at least 9

Summary

Introduction

Breast cancer is the most commonly diagnosed cancer and the leading cause of cancer death among females in the world based on the GLOBOCAN 2018 estimates of cancer incidence and mortality [1]. Diffusion tensor imaging (DTI), an extension of diffusion weighted imaging (DWI), has been used to characterize breast lesions and shows promising results in increasing diagnostic specificity [4]. It can calculate the anisotropy and directionality of water diffusion in tissues by encoding the diffusion in six or more directions [5]. Several studies indicated that DWI-derived apparent diffusion coefficient (ADC) values, named MD in DTI protocol, significantly decreased in breast cancers compared with benign lesions, and it increased the ability of DCE-MRI to differentiate cancers from benign lesions [4, 6].

Methods

Results

Conclusion