To evaluate the ability of diffusion tensor imaging (DTI) to quantitatively determine response of breast cancer to neoadjuvant chemotherapy (NAC), and provide preoperative assessment of residual tumor extent before surgery.Breast dynamic contrast-enhanced (DCE) MRI has been known as a useful adjunct modality in monitoring patients treated with NAC ¹. There is also growing evidence on DWI usefulness in evaluating breast cancer early response to NAC ². We have recently demonstrated that the parameters derived from DTI show high ability to diagnose breast cancer ³. Herein, we present a preliminary prospective study with standardized blinded analysis, aimed in evaluating the ability of DTI parameters to monitor breast cancer response to NAC, as compared with DCE-parametric maps and histopathology.

Twenty patients treated with NAC (FEC and Docetaxel) were studied before and after treatment. Images were acquired on a 1.5 Tesla Intera Achieva scanner (Philips). The MRI protocol included axial DTI with fat-suppression (resolution:2.08x2.08x2.5mm, TE/TR of 71msec/14120msec, b values 0 and 700 sec/mm² and 15 diffusion gradient directions), and a DCE protocol (TE/TR/flip angle = 3.45msec/5.58msec/20°, 0.72x0.72x2.0mm resolution, 1-pre and 6 post-contrast time points, within a total of 10.5min).

The DTI and DCE datasets were analyzed separately by two readers who were blind to the clinical and pathological reports. DTI datasets were analyzed using a propriety software, yielding three eigenvectors and their corresponding eigenvalues λ1, λ2 and λ3, their average, and two anisotropy indices⁵. Tumors’ ROI were delineated on λ1 maps, using a threshold ≤1.7x10^-3 mm²/sec, with the aid of the pre-treatment post contrast enhancement images. DCE-images were analyzed with the 3TP method, as previously described⁴. Tumor diameter and volume were determined separately on the λ1 and 3TP parametric maps, and were compared with the pathologic results using Miller&Payne pathologic evaluation criteria (M&P) as a gold standard ⁵.

Response to therapy was determined by measuring changes in diameter and volume of the tumors. Changes in the values of the DTI parameters and patterns of enhancement also served to indicate response (Figure 1) . The median diameter of the lesions pre-treatment was 30.4 mm (range: 12.6-77.9mm) on λ1 maps and 31.4mm (range: 12.6-76.5mm) on the 3TP parametric maps.

Of the twenty patients, eight showed no response (<30% reduction), seven showed minor partial response (PR) (30%-90% reduction), one patient showed major PR (>90% reduction), and four patients showed complete response. DTI Response to therapy was associated with an increase in λ1, λ2, λ3, and in maximal anisotropy but not in FA, DCE response was associated with a decrease in initial enhancement rate and a change towards a delayed wash-out pattern. Pre-treatment and post-treatment/ preoperative assessment of tumor extent revealed a high congruence between the λ1 maps derived from DTI and the 3TP maps (Figure 1). Indeed, Pearson correlation factors between the two methods were of 0.88 and 0.98 for lesion diameter, and volume, respectively.

Analysis of the λ1 maps predicted absence or presence of response when comparing to M&P in 16/20 lesions on the basis of % change in diameter and in 19/20 lesions on the basis of % change in volume. Predicting exact type of response (no response, minor, major, or complete response) was in 13/20 (diameter) and in 15/20 (volume) patients. Analysis of the DCE datasets predicted absence or presence of response in 17/20 lesions on the basis of % change in diameter and in 14/20 of lesions on the basis of % change in volume. Predicting exact type of response was in 14/20 (diameter) and in 11/20 (volume) patients. The same Pearson correlation factor of 0.82 was obtained for the %change in tumor diameter and tumor volume between analysis based on λ1 maps and on 3TP maps, indicating a similar ability of these two methods to estimate response to NAC. Acceptable Pearson correlation factors of 0.67 and 0.62 were obtained between post-treatment pathology diameter and the diameter based on DTI and DCE-MRI, respectively.

We have shown that DTI, due to its dependency on structural and physiological features of the extra and intracellular tissue compartments, primarily cellular density, enables evaluating response to NAC, in a quantitative manner. A high similarity in response assessment was found between the DTI and DCE methods with a good agreement of both with the pathology results. Applying DTI as a routine protocol for monitoring response to NAC is feasible and has a potential to become a valuable method, with the advantage of avoiding the use of injecting contrast agents. However, improvements are required to overcome technical problems⁶ and to increase the accuracy and the spatial resolution of DTI.