To our knowledge, this is the first study utilizing magnetic resonance elastography to non-invasively assess the mechanical properties of pediatric low-grade gliomas. We found that gliomas are significantly softer, and significantly less viscous (lower damping ratio) than normal appearing white matter. We found that the degree of tumor softness appears to be unrelated to tumor size or location, though larger tumors typically show more mechanical heterogeneity than smaller tumors. Interestingly, we also found that in patients with neurofibromatosis type 1, areas of focal abnormal signal intensity have lower damping ratio than the normal appearing white matter.

Our results agree with previous MRE findings in adults, where gliomas are found to be softer than surrounding tissue. Although adult tumor mechanics have been studied, pediatric tumors cannot be assumed to present with the same behavior or hold the same implications without dedicated pediatric research studies. Pediatric gliomas have distinctly different cellular and molecular etiologies than their adult counterparts, leading to differences in tumor location, tumor burden, tumor growth trajectories, and prognoses. In pediatrics, it can be challenging to assess tumor malignancy using a traditional Ki-67 immunostaining, as the developing central nervous system has a naturally robust proliferative potential [3]. Further, glial progenitor cells which serve as precursors for tumors are found in variable abundances in children and adults, and their rates of proliferation vary significantly [25, 26]. These differences in pediatric tumor make it necessary to studying pediatric tumors separately.

In this work we chose study LGGs, as opposed to the more often studied high-grade gliomas, as treatment for LGGs including chemotherapy and radiation can interfere with developmental processes [27]. In children, the decision to intervene in high-grade gliomas is much clearer due to the significant, imminent threat they pose, whereas in low-grade tumors, if delaying treatment is possible, it may help avoid affects to normal neurodevelopment [27]. Here we find that LGGs are on average ~10% softer than normal appearing white matter. Our study only focuses on LGGs, and therefore we cannot explicitly comment how pediatric glioma mechanical properties are associated with tumor grade, though based on past work, we expect that lower tumor stiffness would occur with increasing grade of pediatric tumors. One adult study showed that Grade II gliomas were ~ 30% softer than reference tissue, while Grade III gliomas are ~ 44% softer than baseline, and Grade IV tumors are ~50% softer than baseline [18]. Our observation of ~10% lower stiffness in Grade I and II tumors in the present study would appear to be consistent with this adult trajectory of grade-dependent softness. Animal models of gliomas have also shown decreasing viscosity across several weeks of tumor progression [28, 29], with a 17% decrease in damping ratio over a four-week period of tumor growth.

It is common practice to consider the mechanical properties of tumors relative to healthy reference tissue, as significant variability in stiffness and damping ratio exists between individuals. Thus, it should not be overlooked that children and adults have fundamental differences in the mechanical properties of their normal brain tissue. Brain maturation between age 5- and 35-years shows decreases in stiffness of approximately 0.3% per year, with a 5-year-old having an average brain stiffness of ~3.2 kPa while a 35-year-old has an average brain stiffness of ~2.9 kPa [30]. Tissue viscosity increases at an approximate rate of 0.4% per year over the same time. While there is no explicit elastography work about the expected alterations to tumor mechanical properties which form in stiffer and softer brains, cytopathology research has repeatedly demonstrated significant differences between the structure and function of cells which were grown on substrates of differing mechanical properties [31].

Despite variations in pediatric and adult cellular markers in tumors, histopathology and immunohistochemistry from adult murine glioma mechanical property studies can allow us to speculate on the implications of our findings [28]. Morphologically, tumors cells are large and amorphous compared to neural cells and have less structural cohesion between cells, making them more soft than healthy neural tissue [32]. Tumors also vary molecularly; one study shows that stiff and soft tumors have different genetic signatures, with genes for matrix reorganization and cellular adhesion being overexpressed in more stiff tumors [33]. More research is needed to fully explain the microstructural changes which portend soft gliomas, but elastography has been demonstrated as a unique tool to investigate these changes [34]. It should be noted, however, that not all brain tumors are soft; for instance, meningiomas are both stiffer and have a higher viscosity than surrounding tissue [13, 28].

In pediatrics, unlike adults, genetic factors such as the presence of NF1 are a primary contributor to the formation of gliomas, and NF1 nearly always shows the presence of non-neoplastic FASIs. As the size and number of FASIs can fluctuate over time, early-stage gliomas can be mistaken for FASIs in patients with NF1. Differentiating FASIs from healthy tissue, and from lesions likely to show radiologic progression, can be the defining factor in developing a suitable tumor monitoring plan. Here, we find that FASIs appear as slightly stiffer that normal appearing reference tissue, unlike gliomas which are soft, though not significantly. This can be attributed to the large range of stiffness variability found in FASIs, and their small nature making them more challenging to isolate. Similar to gliomas, FASIs have a significantly lower damping ratio than reference tissue. To our knowledge, this is the first time the mechanical properties of non-neoplastic FASIs in NF1 have ever been measured in vivo.

The mechanical underpinnings of FASI pathology are not well understood. FASIs show heterotopia, increased size and number of myelin vacuoles, gliosis, a buildup of intramyelinic edema, and macrocalcification foci [35]. Together these abnormal microstructural variations appear to have a combined biomechanical effect, and while these phenomena have not been individually assessed to determine their contributions to FASI mechanics, we can speculate on their individual contributions. MRE has shown that in hydrocephalus, which is characterized by a presence of unbounded fluid, brain tissue appears soft [36]. Conversely, bounded intramyelinic edema may be stiffening the individual myelin sheaths due to swelling, which in aggregate could make FASI areas stiffer. Macrocalcification foci refer to larger or more prominent accumulations of calcium; in most organs, calcium deposits lead to a macroscale stiffening of the tissue area [37]. Finally, gliosis refers to the proliferation and hypertrophy of glial cells, which occurs usually in response to inflammation; gliosis can form glial scars which are a dense network of astrocytic tissue [38]. However, increasing size and number of myelin vacuoles is likely to reduce mechanical integrity [39], and in some cases glial scars have shown to be mechanically soft [40]. Therefore, these contradictory processes make it challenging to elaborate on the expected directionality of mechanical properties of FASIs without more in vivo and ex vivo work.

This study had several limitations. Gliomas are heterogenous, and even with restricting our sample to LGGs, we observed large variability in tumor size, location, and cell type in our population. While we did not see evidence of trends in tumor mechanical properties based on these characteristics, they likely result in some variability not fully characterized. Histopathological diagnoses were not available for all the included tumors as some LGGs were never resected. Further, our population spans the pediatric age range from age four and older, and brain tissue changes significantly during maturation. To account for this, we used an individual’s normal appearing white matter as the reference, but future studies with larger samples sizes could more appropriately parse the interaction between maturation and tumor properties. Finally, image quality and resolution could have influenced our findings. We confirmed quality of the shear wave using the standard MRE metric octahedral-shear strain signal-to-noise ratio, but it is possible that other image artifacts could influence our data, including geometric distortion near high susceptibility regions [41]. We also acknowledge that the images were lower resolution than optimal for measuring the FASIs, which can be smaller than 1 mm. Because they are often clustered close together, we were able to treat them as a distinct unit, but their size created limitations for individual FASI analysis.