Background: American College of Radiology Imaging Network (ACRIN) trial 6657, the imaging component of the I-SPY trial (CALGB 150007/150012), is testing MRI for predicting response to treatment and stratifying risk-of-recurrence in patients with locally-advanced breast cancer. We report preliminary results evaluating MRI for prediction of pathologic response. 

Methods: Women with ≥3 cm invasive breast cancer receiving neoadjuvant chemotherapy (NACT) with anthracycline-cyclophosphamide (AC) followed by a taxane (T) were enrolled from May 2002 to March 2006. MRI was performed prior to NACT (t1), after 1 cycle AC (t2), between AC and T (t3), and following T prior to surgery (t4). MRI tumor size assessments included longest diameter (MRLD) and tumor volume (MRVol). Clinical size (clinsize) and mammographic longest diameter (MGLD) were also recorded. Linear dimension was measured by the radiologist for MGLD and MRLD; MRVol was calculated by computer using signal enhancement ratio (SER) thresholds. Change in clinical and MRI variables at t2 were compared for ability to predict pathologic complete response (pCR). 

Results: 237 patients were enrolled at 9 institutions. 216 patients with complete imaging were analyzed. Of tumor size measurements at t4, MRVol showed the strongest correlation with pathsize among clinsize (r = 0.44), MGLD (ns), MRLD (r = 0.28) and MRVol (r = 0.61). Early change in MRVol measured at t2 was the only variable predictive of pCR among clinsize (p = 0.14, 0.15), MRLD (p = 0.40, 0.07), MRVol (p = 0.02, 0.01) and peak SER (p = 0.53, 0.72) in univariate and multivariate logistic regression, respectively. 

Conclusions:Tumor response measured volumetrically by MRI is a stronger and earlier predictor of pathologic response after NACT than clinical exam or tumor diameter. This work is funded by NIH/ACRIN U01 CA79778; CALGB CA31946, CA33601; NCI SPORE CA58207.