Paper Info
Title
Predicting response before initiation of neoadjuvant chemotherapy in breast cancer using new methods for the analysis of dynamic contrast enhanced MRI (DCE MRI) data
Abstract
The pharmacokinetic parameters derived from dynamic contrast enhanced (DCE) MRI have shown promise as biomarkers for tumor response to therapy. However, standard methods of analyzing DCE MRI data (Tofts model) require high temporal resolution, high signal-to-noise ratio (SNR), and the Arterial Input Function (AIF). Such models produce reliable biomarkers of response only when a therapy has a large effect on the parameters. We recently reported a method that solves the limitations, the Linear Reference Region Model (LRRM) Similar to other reference region models, the LRRM needs no AIF. Additionally, the LRRM is more accurate and precise than standard methods at low SNR and slow temporal resolution, suggesting LRRM-derived biomarkers could be better predictors. Here, the LRRM, Non-linear Reference Region Model (NRRM), Linear Tofts model (LTM), and Non-linear Tofts Model (NLTM) were used to estimate the R-Ktrans between muscle and tumor (or the K-trans for Tofts) and the tumor k(ep,TOI) for 39 breast cancer patients who received neoadjuvant chemotherapy (NAC). These parameters and the receptor statuses of each patient were used to construct cross-validated predictive models to classify patients as complete pathological responders (pCR) or non-complete pathological responders (non-pCR) to NAC. Model performance was evaluated using area under the ROC curve (AUC). The AUC for receptor status alone was 0.62, while the best performance using predictors from the LRRM, NRRM, LTM, and NLTM were AUCs of 0.79, 0.55, 0.60, and 0.59 respectively. This suggests that the LRRM can be used to predict response to NAC in breast cancer.
Year
DOI
Venue
2016
10.1117/12.2217008
Proceedings of SPIE
Keywords
Field
DocType
DCE,MRI,pharmacokinetic,predictive modeling
Breast cancer,Chemotherapy,Arterial input function,Biomarker (medicine),Artificial intelligence,Oncology,Computer vision,Internal medicine,Medical physics,Area under the roc curve,Dynamic contrast-enhanced MRI,Physics,Magnetic resonance imaging
Conference
Volume
ISSN
Citations 
9788
0277-786X
0
PageRank 
References 
Authors
0.34
0
6
Name
Order
Citations
PageRank
Joseph B. DeGrandchamp100.34
Jennifer G. Whisenant200.34
Lori R. Arlinghaus3113.35
Vandana G. Abramson400.34
Thomas E. Yankeelov5207.14
Julio Cárdenas-Ródriguez600.68