(P 329) DEVELOPING ARMS METASTATIC MODELS TO RECAPITULATE THE TUMOR MICROENVIRONMENTS OF THE AXILLA/LYMPH NODE AND LUNGS

Objective: Alveolar rhabdomyosarcoma (ARMS) is a skeletal muscle soft tissue sarcoma driven by a PAX3/7-FOXO1 (P3F) fusion and is the most aggressive pediatric RMS subtype. Most metastatic patients develop resistance to standard therapies and tumor microenvironment (TME) heterogeneity can significantly affect tumor progression and treatment response. To improve drug testing platforms and patient stratification, there is a need for preclinical models representing specific metastatic niches. The lungs are the most common metastatic site for ARMS while the axilla/lymph nodes (ALN) are particularly aggressive. Our goal is to develop reproducible and quantifiable ARMS metastatic models of the lungs and/or lymph nodes.

Methods: P3F fusion-positive human RH4 and RH30 cell lines were obtained from ATCC, the murine M53HP3F cell line was created by the Gladdy lab and dissociated ARMS PDXs were provided by PROFYLE. Using a Pediatric Cancer Atlas including an ARMS cluster and the OTTER ensemble system, we compared the cell lines’ transcriptomic signatures with these clusters. Invasion was assessed using a Matrigel-coated transwell assay, while migration was assessed using uncoated inserts. Cells were fixed with PFA, stained with Hoechst, and imaged using fluorescent microscopy. To establish surgical metastatic models, NSG mice (n=10) were intramuscularly injected with luciferase-expressing ARMS cells, followed by hindlimb amputation. Metastases (mets) were detected using bioluminescent imaging (BLI), MRI and caliper measurements; followed by necropsy. ALN metastatic histology was evaluated by a sarcoma pathologist, and P3F fusion was assessed using RT-PCR. An ex vivo pulmonary lung metastasis assay (PuMA) was established by injecting GFP-expressing RH4 cells into the mouse tail vein (n=4), followed by euthanasia. The lungs were insufflated with agarose, solidified, sliced, and incubated in media. Lung metastatic burden was monitored with confocal fluorescent microscopy and analyzed via ImageJ.

Results: Using surgical methods, the cell lines and PDXs consistently developed metastases at sites relevant to human cancer including the ALN, liver and retroperitoneum. Metastatic tumors displayed ARMS histopathology and preserved the P3F fusion. The transcriptomic profile of RH4 clustered closely with the P3F fusion-positive ARMS signature from the cancer atlas, indicating RH4’s accurate representation of human disease. Thus, RH4 was prioritized to validate BLI as a quantification tool in this model. RH4 exhibited migration and invasion capacity in vitro with a stronger propensity for migration. In the RH4 surgical model, ALN metastases occurred with the highest consistency in 90% of the cohort across different cell line passages. Primary xenografts had a latency of ~3.6 weeks, and ALN metastases became palpable ~33 days post-op. BLI detected ALN metastases with high sensitivity, ~20 days post-op, and the signal intensity correlated with tumor size (Fig.1). MRI was used to confirm the presence and size of the tumor. Given that lungs are the most common metastatic site, we developed an ARMS PuMA model which showed a gradual increase in lung metastatic burden over time: 4-fold increase (p< 0.05) by day 4 and 8-fold increase (p< 0.0001) by day 6 normalized to day 0 (Fig.2).

Conclusion: Our metastatic models demonstrate clinical relevance and reproducibility, offering platforms for quantitative tumor burden monitoring that will be valuable for drug testing. The latency of these models makes them highly efficient for preclinical work. The ALN model is ideal for in vivo studies, while PuMA is suitable for drug testing using in vitro conditions while recapitulating the lung TME. Future research studies will focus on molecularly characterizing the PDX cells that metastasize to the ALN to identify biomarkers that drive migration of the majority of ARMS tumors to this site.