Development of a murine leptomeningeal disease model for a point-of-care cerebrospinal fluid liquid biopsy assay

Background: Leptomeningeal disease (LMD), the metastatic dissemination of cancer cells to the leptomeninges and cerebrospinal fluid (CSF), is a devastating complication of cancer. LMD occurs most commonly in advanced melanoma, lung, and breast cancers, and typically portends a poor prognosis that is accompanied by rapid neurological decline. Diagnosis and disease monitoring has been historically challenging. Prior work by our colleagues has led to the development of a barcode-style rapid, semi-quantitative lateral flow assay (LFA) able to detect MCF-7 breast cancer cells in artificial CSF using an epithelial marker. Here, we present murine models of LMD developed to further refine the assay and test its detection limit in an in vivo setting. Methods: Two brain-tropic murine cancer lines previously generated by serial in vivo intracarotid injections, B16-BrM melanoma and LLC-LeptoM lung carcinoma, were engineered to stably express mCherry and secreted Gaussia luciferase (sGluc) via lentiviral transduction. Pure populations were isolated by fluorescence-activated cell sorting, and bioluminescence signal was evaluated in vitro using cultured supernatant at various cell concentrations. For in vivo studies, 1000 parental or mCherry-sGluc cells were stereotaxically injected into the right lateral ventricle of syngeneic C57BL/6 mice at 8-10 weeks of age. CSF was collected terminally via cisterna magna puncture on days 1, 5, 9, and 13 post-implantation. Tumor burden was assessed using bioluminescence and flow cytometry on CSF. Murine brains were collected and processed for histological analyses. Results: Bioluminescence signal was detectable in cultured supernatant from both B16-BrM and LLC-LeptoM cells, even in concentrations less than 10 cells per ml. When adopted into the in vivo LMD model, mice began to display clinical decline starting around day 12 post-implantation. Histological analysis demonstrated tumor cell clusters sequestered in the ventricular space. Bioluminescence signal was detectable from just 1 ul of mouse CSF in animals injected with mCherry-sGluc cells, compared to those injected with parental cells or with no injection. On day 13 post-implantation, bioluminescence signal was saturated in about half of the experimental cohort. In most cases, bioluminescence was detected in CSF even when mCherry+ cells were not detected on flow cytometry. Conclusion: We have established a murine LMD model that allows for highly sensitive detection of tumor burden via liquid biopsy of CSF. Further studies evaluating tumor burden kinetics and the performance of the point-of-care LFA on mouse and human CSF are ongoing. The success of this work may revolutionize the diagnosis and management of LMD.