Patient stratification is essential for addressing breast cancer's inter- and intra-heterogeneity. While the PAM50 classifier is widely regarded to be the gold standard for breast cancer molecular subtyping, there are challenges associated with this method including high costs and potential delays in treatment delivery. To address these challenges, we implemented a graph-based multimodal model that integrates histopathology whole slide images and bulk RNA sequencing data. This approach assesses whether we can use histopathology slide images alone (or in combination with other modalities) to improve the clinical utility of molecular subtyping. By incorporating RNA sequencing, the model identifies a minimal gene set that can be used in addition to image analysis for molecular subtyping. By doing so, we can evaluate whether multimodal approaches can enhance the efficiency and accessibility of breast cancer treatment delivery.