Automated tracking of nanoparticle-labeled melanoma cells improves the predictive power of a brain metastasis model

Biologic and therapeutic advances in melanoma brain metastasis are hampered by the paucity of reproducible and predictive animal models. In this work, we developed a robust model of brain metastasis that empowers quantitative tracking of cellular dissemination and tumor progression. Human melanoma cells labeled with superparamagnetic iron oxide nanoparticles (SPION) were injected into the left cardiac ventricle of mice and visualized by MRI. We showed that SPION exposure did not affect viability, growth, or migration in multiple cell lines across several in vitro assays. Moreover, labeling did not impose changes in cell-cycle distribution or apoptosis. In vivo, several SPION-positive cell lines displayed similar cerebral imaging and histologic features. MRI-based automated quantification of labeled cells in the brain showed a sigmoid association between metastasis frequency and doses of inoculated cells. Validation of this fully automated quantification showed a strong correlation with manual signal registration (r²=0.921, P < 0.001) and incidence of brain metastases (r2= 0.708, P < 0.001). Metastasis formation resembled the pattern seen in humans and was unaffected by SPION labeling (histology; tumor count, P=0.686; survival, P=0.547). In summary, we present here a highly reproducible animal model that can improve the predictive value of mechanistic and therapeutic studies of melanoma brain metastasis. Cancer Res; 73(8); 2445-56.