Invasive lobular carcinoma (ILC) accounts for 15% of breast cancers yet lacks specific therapy because ILCs are underrepresented in clinical trials and preclinical models are lacking. Here, we established intraductal xenograft models to test whether the clinical pan-lysyl-oxidase inhibitor PXS-5505, now in phase I/IIa trials for myelofibrosis, can exploit the collagen-rich matrix dependency in ILC created by CDH1 loss. PXS-5505 remodeled fibrillar collagen and halted tumor expansion and metastatic seeding across ER+ and triple negative models without systemic toxicity. Genome-wide CRISPR screens revealed ITGAV and ITGB5 as synthetic lethal partners of CDH1, and LOX inhibition downregulated their expression, together with MYC, NF-κB, and AP-1 transcriptional programs. Collagen fiber density/alignment and MYC/AP-1 gene signatures served as pharmacodynamic readouts of drug activity. These data uncover a tractable ECM-integrin-MYC axis in ILC and nominate PXS-5505, alone or with endocrine therapy, for window of opportunity trials in this understudied breast cancer subtype.