Diffuse midline gliomas (DMG) are a highly aggressive subtype of pediatric high-grade glioma. These tumors are universally fatal, with a median overall survival of less than 1 year. They account for 10-20% of pediatric central nervous system (CNS) malignancies and are the leading cause of pediatric CNS tumor-related death (1).

Focal radiation therapy (RT) remains the only therapeutic approach that has been shown definitively to lengthen DMG survival, although it usually only delays progression of the tumor by a few months. There has been minimal improvement in the prognosis of DMG due to the failure of previous clinical trials (2). Focal RT has been the standard RT field because focal progression in the pons or other primary site is usually responsible for death. However, DMG demonstrates high rates of leptomeningeal dissemination at diagnosis, during treatment, and at autopsy (3, 4). With recent or current early trials of techniques allowing localized, primary tumor-directed delivery of chemotherapy agents, including convection-enhanced delivery (CED) (5) and focused ultrasound with microbubbles (6), the potential for prolonged local control of DMG is becoming more realistic; however, some patients are now experiencing metastatic progression while their primary tumor is responding to these treatments. Thus, controlling metastatic DMG has become a pressing challenge in the fight to achieve long-term survival in this disease. Craniospinal irradiation (CSI) is a sensible modality to consider, since DMG is sensitive to RT, and CSI is crucial to survival in multiple other pediatric brain tumors with metastatic potential, such as medulloblastoma and atypical teratoid/rhabdoid tumor.

In the present study, we aimed to investigate CSI as a therapeutic approach to control metastatic disease across the neuraxis in an orthotopic patient-derived xenograft (PDX) model of DMG. We hypothesized that we could feasibly study DMG preclinically using a PDX model, and that CSI would limit metastatic disease in this model compared to focal RT, although it would not extend survival alone.

Using a murine orthotopic xenograft model of DMG, we identified CSI as a therapeutic modality capable of limiting metastatic disease throughout the CNS compared to focal RT, the current standard-of-care therapy in DMG. This brief report represents the first preclinical assessment of CSI in DMG and raises this treatment modality as a potential contributor to a combination approach in this disease. This approach is necessitated by a developing “good” problem in DMG, that more prolonged local tumor control is potentially becoming more feasible with the introduction of experimental treatment modalities allowing focal delivery of chemotherapy, leading to the emergence of clinically apparent disease dissemination in some patients as their survival has been prolonged.

In our study, mice in the focal RT and no RT groups showed similar levels of metastatic disease. Since the only difference between these groups was pontine RT, this suggests that metastatic tumor was seeded throughout the CNS early in the disease process instead of continuing to disseminate from the primary tumor, as in this case we would expect decreased metastatic disease with focal RT. BLI tended to show spinal metastases only later, however, which matches what has been seen in patients: metastatic DMG is less frequent by MRI at diagnosis than later in treatment (3) but can often be detected at diagnosis by more sensitive assays like circulating tumor DNA, the use of which is experimental but emerging in this disease (4). The timing of treatment in this model, at first detection of abnormal signal by imaging, is necessary in mice given the need for sacrifice at symptom development, but it presents a potential issue in the translation of our findings to patients, who do not present until developing symptoms. However, the findings from our work and human studies (4), both showing metastatic tumor present from first detection of the primary tumor, suggest that CSI delivered at the time of diagnosis would be effective.

Weaknesses of the current study include the use of only one radiation dosing schedule and one model, which harbors both H3K27M and TP53 somatic mutations; TP53 mutation is actually associated with radiation resistance in DMG (10), so we might expect an even more profound impact of radiation in TP53-wild type tumors, and future work on CSI should explore a variety of DMG models. Another weakness is the inability to be sure that the pattern of metastatic disease development in this model matches that seen in patients. However, the injection of cells directly into the pons (as opposed to the ventricle), the variability of metastatic disease development, and the findings on histopathology all suggest accurate modeling of true metastases akin to those seen in patients. Strengths of the study include the multiple assessments of metastatic disease by dual imaging modalities and histopathology, the grading of metastases by a neuropathologist blinded to treatment group, and the novel approach of delivering CSI to mice through a custom-made platform to minimize toxicity.

As expected, CSI did not produce a survival benefit over focal RT in our study, as mice still died of their primary tumor with radiation alone. This again matches what has been seen in patients: RT alone is a palliative measure only in DMG that can only delay tumor progression, including in the few documented patients in which CSI has been given for metastatic tumor (11, 12), especially since even maximal CSI doses are lower than doses that can be delivered to the primary tumor. However, these studies did indicate temporary efficacy against these metastases, and combined with our findings, suggest that CSI has potential as a component of combination, multimodality therapy against DMG to control the tumor dissemination likely to become clinically apparent as treatment measures directed at the focal tumor become more effective at prolonged local control. Our methods and findings provide the groundwork for CSI to be incorporated into future preclinical studies and clinical trials. Combination approaches (13) addressing both the primary tumor through focal RT and targeted delivery of chemotherapy and/or immunotherapy (14, 15), as well as disseminated tumor through CSI and systemic chemotherapy and/or immunotherapy, have the potential to produce long-term control of DMG.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

The animal study was reviewed and approved by Institutional Animal Care and Use Committee, University of Colorado Anschutz Medical Campus.

ALG conceptualized and planned the project, with contributions from RV and KD. ALG, AK, BC, AO, and SK created the radiation plan. AK, BVC, and AO carried out the experiments. NJS led animal imaging (BLI and MRI). AG led histopathological analysis. AK, AG, and ALG led data analysis, assisted by JD, PF, RL, and HC. EB, AK, and ALG wrote the manuscript. All authors contributed to the article and approved the submitted version.