Background: Diffusion-weighted (DW) MRI is a non-contrast technique that can reflect treatment-induced alterations in tumor microstructure and cellularity. ACRIN 6698 was performed as a sub-study of the I-SPY 2 TRIAL to evaluate quantitative DW MRI for early assessment of breast cancer response to neoadjuvant chemotherapy (NAC) in a multisite, multiplatform trial.
Methods:The IRB-approved trial was performed at ten institutions. Of 406 enrolled breast cancer patients, 272 were randomized to treatment (12 weekly cycles paclitaxel+/-experimental agent, followed by AC) and underwent breast DW MRI at four time points: pre-NAC (T1), early-NAC after 3 cycles paclitaxel (T2), mid-NAC between paclitaxel and AC (T3) and post-NAC (T4). Tumor apparent diffusion coefficient (ADC) was measured at each time point and compared for patients with and without complete pathologic response (pCR) by Wilcoxon signed rank test. Exploratory analyses were performed across subtypes defined by hormone receptor (HR) and HER2 expression. Performance for predicting pCR was assessed by calculating the area under the ROC curve (AUC).
Results:Of 272 treated patients, 227 comprised the final cohort (14 were excluded for missing MRI exams, 31 for poor image quality). Median patient age was 48 (range, 25-77) years, and 71/227 (31.3%) achieved pCR. Subtype groups were HR+/HER2+ (n = 38), HR+/HER2- (n = 95), HR-/HER2+ (n = 20), and HR-/HER2- (n = 74). For the full cohort (all subtypes and treatments), both ADC and change in ADC from T1 were significantly predictive of pCR at T3 (AUC = 0.63, 95% CI 0.55-0.71; AUC = 0.62, 95% CI 0.53-0.70, respectively), and also at T4. ADC measures were not predictive of pCR at either T1or T2. Stratifying by subtype showed change in ADC at T3 was more predictive in HR-/HER2+ (AUC = 0.86) and HR+/HER2- (AUC = 0.75) tumors than HR-/HER2- and HR+/HER2+ tumors (AUC = 0.59 and 0.56, respectively).
Conclusions: DW MRI reflects cytotoxic effects of chemotherapy, and mid-treatment ADC was a predictive marker of pCR. The predictive value of ADC varied across biologic subtypes. Further work is needed to determine the comparative predictive value of ADC to other imaging metrics.