Immunomodulatory mechanisms of endoplasmic reticulum stress in the tumor immune microenvironment and prediction of treatment response in HER2-positive breast cancer

Abstract HER2-positive breast cancer (BC) is an aggressive subtype with poor long-term outcomes in advanced disease, largely due to resistance to HER2-targeted therapies. Within the tumor-immune microenvironment (TIME), tumor and immune cells face diverse stressors, including endoplasmic reticulum stress (ERS). ERS activates the Unfolded Protein Response (UPR), which aims to restore homeostasis or trigger apoptosis. In BC cells, chronic UPR activation promotes malignant progression, while in immune cells ERS impairs function and weakens antitumor immunity. Given the relatively high immune infiltration in HER2-positive BC, preserving immune competence is crucial for therapeutic efficacy. This study aims to elucidate TIME-associated molecular mechanisms in HER2-positive BC using in-vitro model and patient tumor samples analysis from 21 patients and to develop predictive models of treatment response using liquid biopsy-based mass spectrometry (MS) approach (21 patients, 15 healthy donors). ERS modulation was shown to significantly affect cancer–immune cell interactions in co-culture models. Key ERS-related genes were characterized at the transcriptional, translational, and spatial levels. These findings were used to model the effects of immunomodulatory therapies in vitro. Analysis of patient samples revealed distinct TIME patterns associated with treatment response. Serum-based MALDI-TOF MS enabled the development of a predictive model capturing systemic alterations associated with ERS driven tumor immune interactions, therefore discriminated responders from non-responders. Modulating ERS within the TIME represents a promising strategy to enhance immune-mediated antitumor activity in HER2-positive BC. This study provides a detailed molecular characterization of the HER2-positive BC TIME and presents an MS- and machine learning-based predictive model for patient stratification.