Prediction of pathologic response to neoadjuvant chemotherapy in patients with breast cancer using diffusion-weighted imaging and MRS

The aim of this study was to determine whether tumor size, MRS parameters and apparent diffusion coefficient (ADC) measurements could be applied to predict pathologic complete response (pCR) after neoadjuvant chemotherapy (NAC). Ninety patients with breast cancer (median size, 4.5 cm; range, 1.6-9.5 cm) were evaluated with single-voxel ¹H MRS and dynamic contrast-enhanced MRI. Diffusion-weighted imaging was performed in 41 of these patients using a 1.5-T scanner before and after completion of NAC. Pre- and post-treatment measurements and changes in tumor size, MRS parameters [absolute and normalized total choline-containing compound (tCho) integral and tCho signal-to-noise ratio (SNR)] and ADCs in pCR versus non-pCR were compared using the nonparametric Mann-Whitney test. Receiver operating characteristic (ROC) curve analysis was performed to assess the diagnostic performance of each parameter. After NAC, 30 patients (33%) showed pCR and 60 (67%) showed non-pCR. At pretreatment, ADC was the only significant parameter in differentiating between pCR and non-pCR [(0.83 ± 0.05) × 10⁻³ versus (0.97 ± 0.14) × 10⁻³ mm²/s] (p = 0.014). Post-treatment measurements after completion of NAC and changes in tumor size (both p < 0.001), MRS parameters (p = 0.027 and p = 0.020 for absolute tCho integral, p = 0.036 and p = 0.023 for normalized tCho integral, and p = 0.032 and p = 0.061 for tCho SNR) and ADC (p = 0.003 and p < 0.001) were significantly different between the pCR and non-pCR groups, except for changes in tCho SNR. In ROC analysis, the areas under the ROC curve (AUCs) of 0.63-0.73 were obtained for tumor size and MRS parameters. AUCs for pre- and post-treatment ADC and changes in ADC were 0.75, 0.80 and 0.96, respectively. The optimal cut-off of the percentage change in ADC for predicting pCR was 40.7%, yielding 100% sensitivity and 91% specificity. Patients with pCR showed significantly lower pretreatment ADCs than those with non-pCR. The change in ADC after NAC was the most accurate predictor of pCR.