As the most prevalent form of primary brain tumors, glioma develops from neural glial cells, mainly star-shaped astrocytes. Based on the classical WHO tumor classification, glioma can be subgrouped into Grade 1–4 (Louis et al., 2016). Despite highly variable histological and genetic characteristics, glioma is notorious for its rapid proliferation, extensive invasion, genetic heterogeneity, and therapeutic resistance. Even given the multiple discipline therapeutics, patients with glioma suffer from the dismal outcomes. Emerging evidence suggests that lncRNAs play a vital role in mediating glioma initiation and progression. Here, we highlighted the current research focusing on the implications of lncRNA on glioblastoma and LGG, respectively (Figure 3).

Pituitary adenomas are mostly located in the anterior lobe of the pituitary gland, which are usually slow-growing and benign. The symptoms of pituitary adenoma may include headache, optic nerve compression, and endocrine changes, including menopause, gigantism, acromegaly, and Cushing’s disease (Bronstein et al., 2011; Holmes, 2016). In 2018, Zhu et al. (2018) suggested that osteoclast differentiation in bone-invasive pituitary adenomas was directly induced by TNFα, which was further regulated by lncRNA SNHG24. Wu et al. (2018) reported that lncRNA H19 inhibited the phosphorylation of 4E-BP1 by preventing the binding of 4E-BP1 to Raptor. In contrast, the overexpression of H19 significantly inhibited the growth of pituitary tumor cells through cell membrane transport. Zhang et al. further demonstrated that the prognosis of patients with prolactinoma was closely related to the expression level of exosomal H19. In addition, the use of cabergoline could enhance the expression of H19 to exert a synergistic therapeutic effect with exosomal H19, which implies the potential of exosomal H19 in the diagnosis and treatment of pituitary tumors (Zhang Y. et al., 2019). Recently, Mao et al. (2020) reported that lncRNA SNHG6 induced the epithelial-mesenchymal transition (EMT) in pituitary adenomas by inhibiting miR-944. Simultaneously, SNHG6/miR-944/RAB11A axis regulated pituitary adenoma proliferation and invasive behavior (Mao et al., 2020).

Meningiomas are the most common benign intracranial tumors arising from the arachnoid cells in the meninges, accounting for 38.3% of all CNS tumors and 54.5% of non-malignant CNS tumors (Louis et al., 2016; Ostrom et al., 2020). The overall survival for patients with benign meningiomas is good, whereas the 5-year survival rate for patients with atypical or malignant meningiomas (WHO grade 2 and 3) is less than 60% (Rohringer et al., 1989). Chromosomal abnormalities at 14q32 are commonly implicated in meningioma pathogenesis and progression (Simon et al., 1995; Weber et al., 1997; Martínez-Glez et al., 2010). MEG3, an imprinted gene located at 14q32, can encode non-coding RNAs with antiproliferative functions (Ghafouri-Fard and Taheri, 2019). Ding et al. (2020) confirmed that lncRNA MEG3 mediated the invasive behavior of meningioma cells through the miR-29c/AKAP12 axis. The upregulation in miR-29c levels can eliminate the adverse effects caused by MEG3 expression on the cell cycle, migration, invasion, and proliferation of meningioma cells (Ding et al., 2020). Zhang et al. (2010) also demonstrated that MEG3 mRNA was highly expressed in normal arachnoid cells but was lost in human meningioma cells, and there was a strong association between the loss of MEG3 expression and tumor grade. Additional evidence has shown that MEG3 could induce the expression of P53, which is a tumor suppressor gene in humans; that is, MEG3 overexpression could reduce the proliferation and metastasis of gastric cancer cells (Wei and Wang, 2017). Alternatively, lncRNA SNHG1 is also associated with meningioma progression. Zhao et al. (2019) reported that SNHG1 was overexpressed in meningioma cell lines, of which deficiency restrained cell growth and accelerated apoptosis. Further mechanism experiments demonstrated that SNHG1/miR-556-5p/TCF12 feedback loop promotes proliferation and inhibits apoptosis in meningiomas through the Wnt signaling pathway (Zhang Y. et al., 2020).

Medulloblastoma is the most common pediatric malignant brain tumor in clinical practice and accounts for 9.2% of pediatric brain tumors (Louis et al., 2007; Millard and De Braganca, 2016; Ostrom et al., 2017). It usually occurs in the lower cerebellar hilum and then metastasizes through CSF circulation (Dufour et al., 2012; Amirjamshidi, 2017). Therefore, CSF cytology has long been one of the routine tests for medulloblastoma. New WHO CNS tumor classification has altered the subgroups of medulloblastomas to mirror new knowledge of their clinical and biological heterogeneity, including 4 principal molecular groups: WNT-activated, sonic hedgehog (SHH)-activated, group 3, and group 4 (Pietsch et al., 2014). SHH is the most common in infants and adults, while other subtypes are common in children (Northcott et al., 2012a, 2017; Taylor et al., 2012). These classifications provide targets for personalized therapy, some of which are currently being tested clinically.

Exploring on the novel molecular biomarkers in medulloblastoma initiation and progression has drawn extensive attention (Ramaswamy and Taylor, 2017). Li B. et al. (2019) reported that lncRNA TP73-AS1 and EIF5A2 were upregulated in medulloblastoma, while miR-494-3p was downregulated. They identify that ELF5A2 is a direct target of miR-494-3p, and knockdown of TP73-AS1 inhibits the proliferation, invasion, and migration of medulloblastoma and promotes cell apoptosis. These findings suggest that lncRNA TP73-AS1 is involved in the development of medulloblastoma as a pro-oncogene (Li B. et al., 2019). Gao et al. (2018a) found that lncRNA LOXL1-AS1 promoted the proliferation and metastasis of medulloblastoma by activating the PI3K/AKT pathway, providing evidence that knockdown of LncRNA LOXL1-AS1 might be a potential therapeutic strategy against medulloblastoma. Moreover, Zhang J. et al. (2020) identified that the expression of lncRNA HOTAIR was higher in medulloblastoma tissues and cell lines than normal samples, which promoted tumor growth, migration, invasion, and EMT by negatively regulating miR-1 and miR-206.

PCNSL refers to the aggressive non-Hodgkin lymphoma in the brain without systemic involvement (Olson et al., 2002; van der Sanden et al., 2002). The most common type of PCNSL is diffuse large B-cell lymphoma (DLBCL), accounting for approximately 2–3% of brain tumors (Langner-Lemercier et al., 2016; Swerdlow et al., 2016; Fox et al., 2019). PCNSL poses an extraordinary challenge to oncologists and neurosurgeons because the impermeability of the BBB hinders the delivery of common chemotherapeutic drugs to the brain (Hanjin et al., 2018). Therefore, intrathecal chemotherapy (injection of chemotherapeutic agents into the CSF via lumbar puncture and delivery of chemotherapeutic agents to the CNS via the cerebrospinal circulation) is reasonable for this malignancy (Cortelazzo et al., 2017).

Recent studies have demonstrated that lncRNAs were widely involved in the biological mechanism of DLBCL by regulating of vital downstream factors through “sponge” intracellular molecules (Huang et al., 2020). For example, lncRNA MALAT1 acts as a ceRNA sponge for miR-195 to stimulate DLBCL cell proliferation and immune escape by activating the immune checkpoint molecules PD-L1 (Wang et al., 2019). LncRNA SNHG14 can act as a ceRNA sponge for miR-5590-3p to upregulate the downstream protein zinc finger E-box binding homeobox 1 (ZEB1). Meanwhile, ZEB1 inversely promotes immune escape of DLBCL cells by transcriptionally activating SNHG14 and PD-L1 (Zhao et al., 2019). SNHG12, another member of the lncRNA SNHG family, can also promote the tumorigenesis of DLCBL by stimulating miR-195 spongiosis (Chen et al., 2020).

Brain metastases refer to the malignant tumors metastasizing to the brain, which are the most common intracranial tumors in adults. Brain metastases may occur up to 10 times more frequently than primary brain tumors (Schouten et al., 2002; Barnholtz-Sloan et al., 2004; Tominaga et al., 2015). Primarily, brain metastases are located in the cerebral hemispheres (80%); the rest were found in the cerebellum and brainstem (Eichler and Loeffler, 2007; Suh et al., 2020). Common cancers causing brain metastases include lung cancer, melanoma, breast cancer, and renal cell carcinoma (Xie et al., 2014). Primary lung cancer is the most common cancer source for brain metastases (Gould, 2018). However, the mechanism underlying the metastatic procedure remains poorly understood. In the cases with multiple metastases, surgery alone usually cannot achieve local control (Mahajan et al., 2017; Churilla et al., 2019). Although whole-brain radiation therapy is acceptable, it is associated with significant cognitive decline (Chang et al., 2009). Similarly, the efficiency of systemic chemotherapy is quite limited, mainly because of the drug resistance and poor penetration through the BBB.

Shen et al. (2015) reported that lncRNA MALAT1 promoted brain metastasis by inducing epithelial-mesenchymal transition in lung cancer, while silencing MALAT1 inhibited highly invasive metastasis cancer cell migration and metastasis by inducing EMT. Furthermore, lncRNA lnc-BM is believed a prognostic indicator for intracranial metastasis patients with breast cancer. Elevated lnc-BM expression promotes STAT3-dependent ICAM1 and CCL2, which mediates communication between breast cancer cells and microenvironment immune cells (Wang et al., 2017). Although the differences in lncRNA expression profiles between primary and metastatic cancer are still controversial, this points to the possibility of distinguishing certain types of metastatic brain tumor with CSF biopsy (Tahira et al., 2011). For instance, Li and colleagues observed the upregulation of lncRNA-CCRR in CSF of metastatic brain tumor from breast cancer.

Clinical diagnosis of brain tumors depends on the evaluation of symptoms and signs, neuroimaging (such as CT, MRI, and PET-CT), and pathological examination of tissues as the gold standard (Di Lullo and Kriegstein, 2017). In this era of precision medicine, non-invasive neuroimaging can hardly provide the necessary molecular profiling of brain tumors or prognostic information. On the other hand, serial samples of brain tumors are difficult to obtain, and, therefore, tracking tumor progression is complex. Moreover, surgical biopsy of the brain tissue is challenging because of the tumor location and hemorrhagic risk. Alternatively, brain tumor biomarkers acquired from the body fluid, including circulating tumor cells, cell-free DNA, and exosomal ncRNAs, have drawn extensive attention to overcome these limitations (Mattox et al., 2019; Le Rhun et al., 2020).

A cancer biomarker can be a substance naturally produced by a tumor or the body’s unique reaction to the presence of diseases (Ariston Gabriel et al., 2020; Simonato et al., 2021). Regarding brain tumors, the presence of BBB hinders the transport of nucleic acids and proteins and dramatically decreases the concentration of tumor biomarkers in the peripheral blood of patients (Antonetti et al., 2021). In contrast, CSF is more closely associated with brain tissue than serum and can be enriched for tumor specific biomarkers (Killer, 2013; Parnetti et al., 2019; Gaetani et al., 2020). LncRNAs in the CSF, therefore, are supposed to be potential candidates as sensitive and accurate early diagnostic and prognostic tools for various brain tumors (Latowska et al., 2020).

An increasing number of studies have evidenced the key roles of lncRNAs in regulating cell proliferation, apoptosis, GSC self-renewal, differentiation, and response to hypoxic stress of different brain tumors. A previous study by Jing et al. (2016) examined the expression of lncRNA CRNDE in 164 gliomas and neighboring non-tumor tissues. Overexpression of CRNDE was correlated with a higher WHO grade, recurrence, and tumor volume expansion in tumor tissues; therefore, elevated expression of this lncRNA may be considered a new prognostic marker in glioma. Similarly, lncRNA HOTAIR has been widely discussed regarding glioma biology. For instance, Suppressing HOTAIR expression inhibits glioma cell proliferation, migration, and invasion, which involves the PI3K/AKT signaling pathway (Ke et al., 2015). Therefore, this lncRNA could be considered a novel prognostic and diagnostic biomarker for glioblastoma (Tan et al., 2018). Furthermore, the expression of lncRNA miR210HG was substantially upregulated in tumor tissue than adjacent normal tissue. Patients with glioma exhibited substantially higher serum miR210HG level than healthy controls, indicating this lncRNA is a potential diagnostic biomarker for glioma (Min et al., 2016). Considering the complicated biological effects and interactions of lncRNAs, the multiple-lncRNA signatures could be better diagnostic and prognostic indicators rather than single lncRNAs. Zhang et al. (2012) suggested that several lncRNAs could be used to distinguish between the stage and type of glioma. This study indicated that approximately 129 lncRNAs were differentially expressed between gliomas and normal brain tissue, demonstrating the capacity of lncRNAs in tumor stratification. Serial studies focused on medulloblastoma have also shown that a variety of lncRNAs, including CCAT1, CRNDE, Linc-NeD125, and PVT1, were associated with tumor progression (Northcott et al., 2012b; Song et al., 2016; Laneve et al., 2017; Gao et al., 2018b). Among these lncRNAs, Linc-NeD125 was overexpressed in medulloblastoma tissues compared to normal brain tissues; further studies showed that its ectopic expression promoted cell proliferation, migration, and invasion of medulloblastoma cells in vitro (Laneve et al., 2017). Interestingly, in a recent report from Li and colleagues, tissue expression and CSF expressions of lncRNA CCRR were both evidently upregulated in breast cancer patients with brain metastases. However, the upregulation of serum lncRNA level was not documented, indicating that CSF lncRNAs might be better biomarkers for intracranial tumors (Li D. et al., 2021). Together, the facts mentioned above indicate that CSF lncRNAs have great potential in diagnosing and predicting brain tumors and provide new approaches for the individualized treatment of patients. More information on lncRNAs as biomarkers for brain tumors is summarized in Table 1.

Although CSF lncRNAs is difficult to be used as a routine screening nowadays, it has exhibited several advantages. First, serial CSF lncRNAs can be acquired by lumber puncture, to monitor the progression in a micro-invasive and dynamic way. Second, CSF lncRNAs are directly secreted and confined within the CNS to eliminate the systemic factors, which helps us to understand the biology and pathophysiology of brain tumors. Third, multiple lncRNA microarray would further improve diagnostic accuracy such as sensitivity and specificity. Fourth, the lowest MRI resolution ranges in the order of millimeters, whereas the dimensions of the tumor cell are in micrometers. Such disparity in scale may lead to delay in diagnosis, which can be compromised by CSF lncRNA biopsy. Finally, lncRNA can be combined with the current neuroimage, rather than replace it. For instance, Wu et al. (2021) observed that lncRNA CASC19 promoted glioma progression by modulating the miR-454-3p/RAB5A axis, which was associated with unfavorable MRI features. Similarly, lncRNA SAMMSON overexpression help distinguishing patients with glioblastoma from diffuse neurosarcoidosis, which shares quite similar radiological features (Xie et al., 2019).

Although conventional strategies for brain tumor treatment have been shown to be promising, it remains a considerable challenge to improve the outcomes of patients. For instance, even given the multidiscipline approaches combined with surgical resection, targeted radiotherapy, high-dose chemotherapy, and novel electric field treatment, the median overall survival for glioblastoma is still less than 21 months (Tan et al., 2020). One of the major obstacles is the poor penetration of BBB. Many attempts have been made to deliver drugs efficiently through BBB (Allhenn et al., 2012), including the intrathecal administration (drug injection to the lumbar arachnoid space) and intraventricular administration (drug injection or infusion into the lateral ventricles of the brain, Calias et al., 2014). Kim et al. (2016) evaluated the efficiency of different anti-miR delivery strategies, including intratumoral, intrathecal, and intraventricular routes, in an orthotopic model of GBM. Intraventricular injection of anti-Let-7 resulted in a significant reduction in target gene expression in the whole tumor, indicating a promising approach for ncRNA therapy in brain tumors (Kim et al., 2016). More recently, Donovan et al. (2020) demonstrated that administration of chimeric antigen receptor T (CAR-T) cells into the CSF could be a highly effective therapy for multiple metastatic mouse models of medulloblastoma and PFA ependymoma. Yang et al. (2019) also reported that intrathecal injection of umbilical cord blood mesenchymal stem cells could improve the pain through lncRNA H19/microRNA-29a-3p/FOS axis. These studies suggest that CSF delivery is practical and promising approach to fight against brain tumors. Specifically, intrathecal and intraventricular administration of lncRNAs packaged by exosomes can be evaluated as novel therapeutics for various brain tumors. For instance, Lai et al. (2014) created a sensitive extracellular vesicle system with high stability and BBB permeability.

Generally, lncRNAs participate in the multiple aspects of tumor biology, such as proliferation, invasion, angiogenesis, treatment resistance, stemness maintenance, and immune suppression. Many ongoing studies, therefore, are designed to identify the lncRNAs with potential anti-tumor characteristics, as summarized in Table 2. For instance, TMZ-based chemotherapy is the fundamental treatment for patients with glioma, especially for malignant glioma. Several lncRNAs have been found to be involved in chemoresistance to TMZ in glioma cells, including lncRNA HOTAIR, H19, and MALAT1 (Jiang et al., 2016; Zhang L. et al., 2020). Recently, Lv et al. (2020) demonstrated that high expression of lncRNA DLEU1 predicted a poor prognosis. Furthermore, silencing lncRNA DLEU1 suppressed TMZ-activated autophagy and promoted the sensitivity of glioma cells to TMZ by triggering apoptosis (Lv et al., 2020). Considering the complicated epigenetic effects and interactions of lncRNAs, single lncRNA can be involved in different tumor biological functions. LncRNA LINC00174 is extensively associated with a variety of cancers, including glioblastoma and LGGs. In particular, LINC00174 facilitates glycolysis and tumor growth by regulating the miR-152-3p/SLC2A1 axis and regulates the miR-138-5p/SOX9 axis to promote chemotherapeutic resistance to temozolomide in glioma (Shi et al., 2019; Li B. et al., 2020). It also acts as an oncogene in glioblastoma via promoting proliferative phenotype (Wang Z. et al., 2020). Besides glioma, lncRNAs can be therapeutic targets in other brain cancers. For instance, knockdown of oncogenic lncRNA CRNDE inhibited tumor development in medulloblastoma cell lines, significantly decreased cell proliferation, and increased apoptosis (Song et al., 2016). Similarly, Xing et al. (2018) indicated that lncRNA LINC00460 promoted MMP-9 expression through targeting miR-539, acting as an oncogenic RNA in the meningioma malignancy and accelerating the proliferation and metastasis of meningioma. In conclusion, these studies suggest CSF lncRNAs have shown great potential as a therapeutic target, although further effort is needed before the clinical application.