Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) is a sequence of MRI scans acquired after the administration of a contrast agent. It is used to capture dynamic changes in tissue enhancement over time, which is shown to be different between benign and malignant lesions. DCE-MRI is a sequence of MRI composted of one pre-contrast and several post-contrast instants. This study leverages machine learning techniques to enhance breast cancer classification using the whole DCE-MRI sequence, addressing the common oversight of underutilizing its temporal dimension. We analyze an in-house dataset, integrating radiomic features from each time instant through (i) feature concatenation and training a Random Forest algorithm (multi-instant RF), and (ii) a graph neural network (GNN) to extract informative embeddings in which nodes correspond to the seven time instants of the DCE-MRI sequence and feature nodes are the radiomic features. Our findings indicate that incorporating temporal information improves classification performance, particularly in reducing false positive rates. Despite the complexity of GNNs, their performance gains are marginal compared to the multi-instant RF, suggesting that shallow models may be equally effective with smaller datasets. Explainable AI methods reveal that the third post-contrast and the pre-contrast instants are most informative for classification, offering new insights for radiology physicians.