Neoadjuvant therapy (NAT) is a standard-of-care treatment modality for breast cancer (BrCa), which consists of the delivery of one or more systemic therapies to reduce the tumor size before surgery. If successful, NAT can achieve the total eradication of BrCa (i.e., pathological complete response, pCR), which correlates with increased survival and the possibility of sparing surgery. While pCR is currently assessed after NAT completion, early predictions of this key pathological outcome during NAT would enable treatment adaptation to maximize treatment-induced tumor elimination and, hence, the chances of achieving pCR. To this end, biomechanistic models of BrCa growth and NAT response informed by routine multiparametric MRI collected before, during, and after treatment have been shown to enable the prediction of pCR status early during NAT. However, existing models often present a simplified representation of the pharmacokinetics (PK) and pharmacodynamics (PD) of NAT drugs. Moreover, state-of-the-art computational approaches are primarily based on finite difference schemes matching MRI voxel volumes, which have limited numerical accuracy. In this talk, we present a biomechanistic model that captures the coupled PK-PD of chemotherapeutic drugs used during NAT of triple-negative BrCa (i.e, a common and aggressive subtype of BrCa). Additionally, we propose a computational approach based on immersed isogeometric analysis via the finite cell method. This numerical scheme enables the definition of the patient-specific geometry as a level set informed by the organ segmentation on T1-weighted MRI within a hexahedral tensor-product background mesh matching the MRI voxel volume, thereby further facilitating local adaptivity around the initial tumor. We show that our model can capture patient-specific BrCa dynamics during NAT and that our computational scheme offers a numerically accurate and efficient approach for personalized tumor forecasting.