Predictive ability of DWI/ADC and DCE-MRI kinetic parameters in differentiating benign from malignant breast lesions and in building a prediction model

Abstract

Objectives: To analyze DWI/ADC and DCE-MRI kinetic parameters on breast MRI between benign and malignant lesions, to obtain independent predictors of malignancy, and to build a possible future predictive model.

Methods: 121 patients (151 breast lesions) were evaluated by 3.0T breast MRI. Based on their post-operative histopathology, we divided them into two groups, benign and malignant. We processed their DCE-MRI images to obtain size of lesions and several kinetic parameters like Time Intensity Curve (TIC), Time To Peak (TTP), Area Under Curve (AUC), Maximum Enhancement, Wash-In (WI), and Wash-Out (WO). We also processed their DWI/ADC maps to obtain their Signal Intensity (SI) ADC values. These parameters were compared between the two groups.

Results: The age of the patients was higher in the malignant group as compared to the benign group (mean=55.9 years and 47.3 years, respectively, P<0.001). Malignant lesions were of greater size than benign ones (mean 2.55 cm and 1.82 cm, respectively, P<0.01). On DCE-MRI, parameters such as TIC, AUC, WI, and WO were statistically significantly different (all P values <0.05). On DWI/ADC, malignant lesions had statistically significant lower ADC values than benign lesions (mean of 1.02x0.001 and 1.72x0.001 mm2/s, respectively, P<0.001). We obtained a cutoff ADC value of 1.18x0.001 mm2/s to differentiate between benign and malignant lesions. Our predictive model results showed ADCmean and TIC as independent predictors of malignancy with an excellent combined model of fit (P=0.000).

Conclusion: Kinetic DCE-MRI and DWI parameters are significantly different between benign and malignant breast lesions. ADCmean and TIC are reliable independent predictors of malignancy in our prediction model, with excellent goodness of fit to predict cancer with a sensitivity of 94.2%.