Tumor-intrinsic transcriptional programs influence engraftment of high-grade serous ovarian carcinoma patient-derived xenografts

High-grade serous ovarian carcinoma (HGSOC) is marked by genomic instability, therapeutic resistance, and frequent relapse. Patient-derived xenograft (PDX) models are widely used in HGSOC research, yet the determinants of successful engraftment and molecular fidelity remain incompletely understood. We established subcutaneous HGSOC PDX models from 30 patients and integrated clinical, genomic, transcriptomic, immune, and geometric network analyses. The engraftment rate was 46.7%. In adjusted models, fresh specimen source and partial chemotherapy response were independently associated with successful engraftment, whereas clinicopathologic features, mutation burden, and canonical driver alterations were not. Baseline expression analysis identified MAPK signaling as the strongest positive pathway-level association with engraftment, although no transcriptomic module remained significant after multiple-testing correction. Geometric network analysis showed the largest nominal baseline edge fraction in DNA damage repair (2/10, 20.0%); however, patient-level DDR curvature did not reach permutation-based significance (Cohen’s d = − 0.76; P = 0.058), and no pathway remained significant after false-discovery correction. Matched biopsy–PDX analyses showed substantial retention of driver mutations but heterogeneous transcriptomic agreement and pathway-level changes after engraftment, with no significant paired network effect after correction. Overall, these findings suggest that tumor-intrinsic transcriptional architecture and therapy-related adaptive states, rather than static genomic alterations, are likely to influence PDX formation and may inform model selection and interpretation of preclinical therapeutic studies.