Improved diagnostic value of whole-lesion histogram and texture analyses on multiparametric breast MRI for papillary neoplasms with non-mass enhancement.

Background: Differentiating between benign and malignant entities remains a complex aspect in the diagnosis of breast papillary neoplasms. This study aimed to assess if analyzing whole-lesion histograms and texture features on multiparametric magnetic resonance imaging (MRI) can enhance the diagnostic accuracy of breast papillary neoplasms presenting as non-mass enhancement (NME).

Methods: In this retrospective analysis, 98 female patients with 98 papillary neoplasms exhibiting NME on dynamic contrast-enhanced (DCE) MRI were enrolled. Two radiologists independently assessed all lesions and later established a consensus on morphological features based on the Breast Imaging Reporting and Data System (BI-RADS) criteria. Quantitative histogram and texture metrics were extracted from four MRI sequences: diffusion-weighted imaging (DWI) with b values of 50 and 1,000 s/mm², apparent diffusion coefficient (ADC) map, and contrast-enhanced T1-weighted subtraction (SUB) magnetic resonance (MR) images. The least absolute shrinkage and selection operator (LASSO) was applied to feature selection. A multivariable logistic regression model was developed using stepwise covariate selection. Diagnostic efficacy was assessed via receiver operating characteristic (ROC) curve analysis.

Results: According to BI-RADS, benign and malignant papillary neoplasms with NME differed significantly in the amount of fibroglandular tissue (FGT), distribution, and time-intensity curve (TIC) pattern (P=0.04, 0.008, <0.001, respectively), yielding an area under the ROC curve (AUC) of 0.792 (sensitivity 67.4%, specificity 84.6%). Quantitative analysis revealed differences in the ADC_{standard deviation (SD)}, ADC_{5th percentile}, ADC_{differential entropy (diff-entropy)}, ADC_contrast, DWI_b50-SD, DWI_b800-mean, and SUB MR_{95th percentile} (P=0.009, 0.01, 0.001, 0.01, 0.001, 0.002, 0.02, respectively), achieving an AUC of 0.908 (sensitivity 82.6%, specificity 88.5%). The AUC of the quantitative model outperformed that of the qualitative model (P<0.001). The AUC of the quantitative model for distinguishing malignant NME papillary neoplasms from benign NME papillary neoplasms in the internal validation set was 0.941, with a sensitivity of 90.4%, and a specificity of 87.0%.

Conclusions: Compared to the qualitative BI-RADS assessment, quantitative analysis of whole-lesion histogram and texture on multiparametric MRI is proven to be more effective in distinguishing between benign and malignant papillary breast neoplasms with NME, in order to avoid overtreatment.