Alina Tudorica1, Karen Y Oh1, Stephen Y-C Chui1, Nicole Roy1, Megan L Troxell1, Arpana Naik1, Kathleen Kemmer1, Yiyi Chen1, Megan L Holtorf1, Aneela Afzal1, Charles S Springer, Jr1, Xin Li1, and Wei Huang1
1Oregon Health & Science University, Portland, OR, United States
Synopsis
DCE-MRI was performed in 28 breast cancer patients (29 tumors) before, during, and after neoadjuvant chemotherapy (NACT). Several DCE-MRI pharmacokinetic (PK) parameters were found to be good early predictors of pathologic complete response (pCR) vs. non-pCR after only one NACT cycle. In addition, several PK parameters and tumor size were significantly correlated with pathologically measured residual cancer burden (RCB).Introduction
Quantitative DCE-MRI has been shown effective for early prediction of breast cancer response to neoadjuvant chemotherapy (NACT) (1,2). Most studies employed the standard Tofts model (TM) for pharmacokinetic (PK) data analysis, and few reported relationships between DCE-MRI PK parameters and post-NACT residual disease, which can have implications for surgical decision making. Here we report preliminary results comparing TM and Shutter-Speed model (SSM) PK analysis of DCE-MRI data, and with imaging tumor size measurement, for evaluation of breast cancer NACT response. The SSM accounts for transcytolemmal water exchange kinetics (3,4).
Methods
28 patients with 29 primary breast tumors (one with 2 tumors) who were treated with NACT consented to research DCE-MRI performed at Visit 1 (V1) - before NACT, V2 - after first NACT cycle, V3 – midpoint (after 3-4 NACT cycles), and V4 - after NACT completion. Axial bilateral DCE-MRI was performed using a 3T Siemens scanner, with 14-20 s temporal resolution and ~ 10 min acquisition time (3,4). Tumor ROI was drawn on post-contrast images and tumor size in the longest diameter (LD) was measured according to the RECIST (5) guidelines. The ROI-averaged and pixel DCE time-course data were subjected to both the TM and SSM analyses to extract Ktrans, ve, kep (= Ktrans/ve), and τi (mean intracellular water lifetime, SSM only) parameters. The whole tumor mean parameter value was calculated as the weighted (by ROI pixel number) average of the single-slice ROI values from all slices covering the entire tumor.
Pathologic response to NACT and residual cancer burden (RCB) were determined from post-NACT resection specimens (6). The pathology endpoints were correlated with the MRI metrics using the univariate logistic regression (ULR) analysis and the Spearman’s correlation (SC).
Results
5 patients achieved pathologic complete response (pCR) while the other 23 (24 tumors) were non-pCRs. Table 1 shows the mean ± SD values of the PK parameters and the percent changes (V21%: percent change of V2 relative to V1) for the two groups, as well as the ULR C statistics values (equivalent to AUC of ROC analysis) for early prediction of pCR vs. non-pCR. Only those with C > 0.8 (indicating good early predictor of response) are listed. V21% and V31% of RECIST LD are listed for comparison. Early changes (V21% or V31%) of several PK parameters, estimated from either model, were good predictors of response, while V21% or V31% of RECIST LD were poor (C < 0.7) early markers of response. Fig. 1 shows representative SSM Ktrans, ve and τi color maps of a pCR (1A) and a non-pCR (1B) at V1 and V2. The RCB can be presented in numerical index values or ranks (I, II, and III) with RCB = 0 for pCR (6). Table 2 lists V4 imaging metrics that discriminated RCB ranks with ULR C > 0.7. V4 Ktrans estimated with either model was a good marker of post-NACT RCB rank. The SC analysis (Fig. 2) reveals positive correlations of V4 Ktrans and RECIST LD and inverse correlation of V4 τi with RCB index value, respectively. These correlations were statistically significant (P < 0.04). Discussion and Conclusion
Consistent with previous studies (2), our preliminary results indicate that changes in tumor microvasculature as measured by DCE-MRI are superior to changes in tumor size for early predictors of NACT response. After only one NACT cycle, the % changes or absolute values of several PK parameters can provide good prediction of eventual pathologic response, implying a potentially important role for DCE-MRI in the emerging era of precision medicine and personalized care. Though tumor size is a poor early predictor of response, it is a valuable marker of RCB. PK parameters derived from either the TM or SSM analysis performed comparably well in early prediction and evaluation of NACT response, suggesting that the differences between the TM and SSM PK parameter values are largely systematic and consequently, do not affect predictive capabilities for therapy response. However, SSM analysis allows quantification of the τi parameter, a potential imaging biomarker of metabolic activity (3,7). The utility of the τi parameter is clearly demonstrated in early prediction of response and assessment of RCB. It is interesting to note that τi is the only pre-NACT imaging metric that can provide good prediction of pathologic response with ULR C = 0.826 (Table 1).
Acknowledgements
Grant Support NIH U01-CA154602References
1. Li et al. Trans Oncol 2014;7:14. 2. Ah-See et al. Clin Cancer Res 2008;14:6580. 3. Springer et al. NMR Biomed 2014;27:760. 4. Huang et al. Trans Oncol 2014;7:153. 5. Therasse et al. J Natl Cancer Inst 2000;92:205. 6. Symmans et al. J Clin Oncol 2007;25:4414. 7. Zhang et al. 2011;101:2833.