Clinical and Molecular Features of Disseminated Pediatric Low Grade Glioma A Systematic Review of Literature

INTRODUCTION Gliomas account for approximately 46% of all pediatric CNS tumors. There is growing awareness of pediatric low-grade gliomas (PLGG) that disseminate to distant parenchymal or leptomeningeal locations either at the time of initial diagnosis or upon disease surveillance. Disseminated PLGGs (dPLGGs) are associated with a poorer prognosis than non-disseminated PLGGs. To date there is no comprehensive report characterizing the genome prole of dPLGGs and their associated management. This systematic review aims to identify the pattern of genetic alterations and treatment outcomes described for dPLGG. METHODS A systematic review of the literature was performed to identify relevant articles. A quality and risk of bias assessment of articles was done using the GRADE framework and ROBINS-I tool, respectively. RESULTS Fifty-two studies published from 1994 to 2020 were included in this review with 368 cases reported. There was sporadic reporting of genetic alterations. The most common genetic alteration observed among study subjects was 1p deletion (76%) and BRAF-KIAA1549 fusion (55%). BRAF p.V600E mutation was found in 7% of subjects. A higher proportion of cases demonstrated primary dissemination compared to secondary dissemination (65% vs 25%). First-line chemotherapy consisted primarily of an alkylation-based regimen and vinca alkaloids. Surgical intervention ranged from biopsy alone to surgical resection and CSF diversion, and depended largely upon tumor location and timing of dissemination. Overall, 73% of cases were alive at last follow-up (median, 40.2 months). All studies reviewed either ranked low or moderate for both quality and risk of bias assessments. While 1p deletion and BRAF-KIAA1549 fusion are the most commonly described molecular alterations in dPLGG, these tumors appear to express heterogeneous molecular and biological characteristics distinct from non-disseminated PLGGs. Additional studies on the molecular and biological features of these tumors are needed to better understand the pathogenesis of dPLGG and to inform the development of additional targeted regimens. A median survival of 48 months (range, 5-175 months) has been reported in literature. 12 Optimum therapy for dPLGG including the role of expectant management is unknown. In most cases, radiotherapy and alkylating-agent-based chemotherapy is used and the utility of targeted molecular therapies remains investigational. 5,9,15,22,23 This systematic review aims to identify the pattern of genetic alterations found in reported cases of dPLGGs, as well as outcomes in relation to adjuvant therapy administered. of dPLGG and non-disseminated PLGG, respectively. 11 Based on the available survival data, 73% of cases reviewed were alive at the last follow-up. Primary disseminated tumors were associated with similar survival rates as secondary disseminated tumors (72% vs. 70%). Survival is likely inuenced by the biologic ramications of molecular alterations specic to each tumor. Tumors expressing BRAF-KIAA1549 fusion have been observed to have a better survival than BRAF p.V600E (69% vs 52%). 7,38 Survival for tumors with wild-type BRAF is better than tumors with BRAF p.V600E but not signicantly different from tumors with BRAF-KIAA1549 fusion. 33,38,58 CDKN2A homozygous deletion is also associated with a worse survival outcome especially when the tumor also harbors BRAF p.V600E . 5,6,32,36,38 From these results, only primary disseminated tumors expressed BRAF- KIAA1549 fusion, which may play a role in the higher survival seen in this cohort. None of the secondary disseminated tumors interrogated for BRAF-KIAA1549 fusion were found to be positive, and two out of the 3 cases tested had BRAF p.V600E. None of the secondary disseminated tumors was interrogated for CDKN2A deletion. While broad generalizations might be gleaned from a review of the literature, the true relationship between molecular alterations and prognosis in dPLGG demands further study.


Introduction
Central nervous system (CNS) tumors are the most frequent solid tumors in children, with a prevalence of approximately 5.6 diagnoses per 100,000. 1 Gliomas of the brain and spinal cord are the most frequent subtype, accounting for approximately 45.7% of all pediatric CNS tumors. 2 Classi ed as World Health Organization (WHO) grade I or grade II malignancies, pediatric-type low-grade gliomas (PLGG) occur more commonly in early childhood compared to highgrade gliomas which are more common in older children. 1 While adult LGGs have a predilection for the cerebral hemispheres and often undergo a malignant transformation, PLGGs can arise throughout the neuro-axis and are less likely to transform. 3,4,5,6,7 The ubiquity of magnetic resonance imaging (MRI) has resulted in an increased detection rate with increased reporting of dissemination of PLGGs throughout the leptomeninges, a phenomenon previously thought to be rare. 8,9,10,11,12 Dissemination can be present at the time of initial diagnosis with or without an identi able primary CNS lesion (primary dPLGG) or at the time of disease progression (secondary dPLGG). 8,9,11,13,14 Recent advances in genetic sequencing and molecular alteration pro ling have led to a better understanding of genetic alterations in PLGGs and has also demonstrated fundamental molecular differences between pediatric and adult low grade gliomas. Commonly identi ed alterations in PLGG include BRAF p.V600E, BRAF duplication, and FGFR alterations. 5,7,15 While isocitrate dehydrogenase (IDH 1/2) mutation is the most common driver in adult low grade glioma, this mutation is rare in PLGG. 16,17,18 Currently, however there is no comprehensive report collating the molecular landscape of dPLGG.
Management of PLGG typically begins with surgical resection or biopsy, depending upon the location and nature of the disease at diagnosis. When possible, gross total resection of PLGG offers the most favorable predictor of long-term outcome with 10-year survival of greater than 90%. 10,19,20 When total resection is not possible or safe, the survival rate is predictably lower (50-85%). 10,14 ,19,20 Disseminated PLGG demonstrates a variable prognosis as some tumors run an indolent clinical course with prolonged progression free survival while others exhibit a very aggressive behavior. 10,12,15,21,22 A median survival of 48 months (range, 5-175 months) has been reported in literature. 12 Optimum therapy for dPLGG including the role of expectant management is unknown. In most cases, radiotherapy and alkylating-agent-based chemotherapy is used and the utility of targeted molecular therapies remains investigational. 5,9,15,22,23 This systematic review aims to identify the pattern of genetic alterations found in reported cases of dPLGGs, as well as outcomes in relation to adjuvant therapy administered.

Methods
A systematic review of the literature was performed in accordance with the Preferred Reporting Items of Systematic Reviews and Meta-analyses (PRISMA) guidelines. 24 . The search was conducted in OVID Medline and PubMed electronic databases in January 2021 to identify relevant articles published between 1990 and 2020. Medical Subject Headings (MeSH) and non-MeSH terms used included "low grade glioma", "disseminated low grade glioma", "diffuse leptomeningeal glioneuronal tumor", "disseminated leptomeningeal tumor", and "leptomeningeal dissemination". After title and abstract review, the articles were exported and managed using EndNote 20. Searches in the databases were supplemented by manual search to retrieve additional articles identi ed via reference list review of the initial set of articles. Article inclusion and exclusion were deliberated among two authors (JHC and MCD).  (Table 2); 36 were unspeci ed. Average time to secondary dissemination was 21.9 months from original tumor diagnosis. There was sporadic testing and reporting of molecular alterations. Thirty-two studies conducted some genetic analysis of tumors including interrogation for BRAF p.V600E, BRAF-KIAA1549 fusion, TP53 mutation, IDH mutation, 1p deletion, 19q deletion, 1p/19q co-deletion, FGFR mutation, and CDKN2A deletion. The remaining 20 studies did not report on molecular alterations. Of the studies which did not report on molecular alterations, 50% were published within the last 10 years. Out of 151 tumors (41%) which were interrogated for some genetic alterations (Fig. 2a), 88 tumors (58%) were found to harbor at least one genetic alteration. The most common genetic alteration observed among all study subjects was 1p deletion (76%, n=64/85) and BRAF-KIAA1549 fusion (55%, n=52/95) ( Fig. 2b and Fig. 2c). BRAF p.V600E was found in 7% (n=5/72) of subjects tested (Fig. 2d). IDH mutation was absent in all 37 cases examined. FGFR1 and CDKN2A analysis was reported by only 2 studies and found to be wild-type in all 10 subjects. Few studies reported genetic alterations speci cally for secondarily disseminated tumors. In this group, 3 subjects were tested for BRAF p.V600E, with 2 being positive. None of these tumors was analyzed for CDKN2A deletion. Only 1 patient with secondary disseminated tumor was tested for 1p deletion and was found positive. All remaining genetic testing among secondarily disseminated tumors was unremarkable ( Table 2).
Rates of tumor biopsy and resection varied widely and depended largely upon location of tumor and timing of dissemination. For cases with primary tumor dissemination who received surgical intervention, 72% had biopsy and 28% underwent resection of primary tumor focus. Among the cohort with secondary tumor dissemination, 35% had undergone biopsy and 65% received an upfront tumor resection (prior to dissemination). One-hundred and two out of 368 (28%) patients required CSF diversion via shunt or ventriculostomy. Adjuvant therapy was reported for 250 cases. Two hundred and eight cases (83%) received chemotherapy and 66 cases (26%) received radiation therapy. First-line chemotherapy consisted primarily of an alkylation-based regimen and vinca alkaloids, with 98 cases (47%) receiving vincristine and carboplatin as rst line chemotherapy. More heterogeneous second-line regimens were reported (Table 3). Three cases (1.4%) received targeted therapies: BRAF or MEK 1/2 inhibitors. Among those who received radiation therapy, 71% received craniospinal radiation and 29% received focal radiation.

Discussion
Despite increasing awareness of dPLGGs, there is limited knowledge on their molecular pro le and long-term response to adjuvant therapy. In this systematic review, we report the molecular alterations, treatment offered and the survival experience of 332 children with dPLGGs reported in the literature. Sixty-ve percent of cases had primary disseminated disease and 25% had a secondary tumor dissemination at an average time of 21.9 months from solitary tumor diagnosis, with timing of dissemination unknown in 47%. Sixty-two percent of studies reviewed characterized some of the molecular alterations harbored by this subgroup of PLGG.
The most common molecular alterations in dPLGG cases identi ed in this review were whole arm chromosomal loss of 1p (64/85) and BRAF-KIAA1549 fusion (52/95). Alterations encountered less frequently were 19q deletion (19/87), 1p/19q co-deletion (18/87) and BRAF p.V600E (5/72). This suggests that 1p deletion may have a mechanistic role in glioma leptomeningeal or parenchymal dissemination. Seventy-four percent of cases reviewed were not interrogated for 1p deletion. Further research is needed to ascertain the possible role of 1p deletion in PLGG dissemination. While the frequency of BRAF-KIAA1549 fusion in non-disseminated PLGGs is high (34 -73%), 1p deletion, either with or without 19q co-deletion, is uncommon (3 -15%) and has mainly been described in oligodendrogliomas, occurring with IDH mutation. 7,32,33,34,35 Frequency of BRAF p.V600E in PLGG differs by histology. 36,37,38 A high rate of mutation is found in pleomorphic xanthoastrocytoma (50 -78%) with moderate rates in gangliogliomas (13 -49%) and lower rates in pilocytic astrocytoma and other glioma subtypes (0 -14.3%). 38,39,40 Fukuoka et al have identi ed a unique IDH wild-type oligodendroglioma-like tumors harboring BRAF p.V600E with no 1p/19q codeletion in a small subset of adolescents and young adults. 41 Other alterations including FGFR mutation or fusion and deletion of CDKN2A are encountered less frequently in PLGGs. 7,36,37,38 The biologic features which permit PLGG to disseminate throughout the craniospinal axis remain unclear and the role of speci c molecular alterations in this process is unknown. Previous studies have suggested tumor dissemination occurs via the CSF pathway with tumor cells penetrating ependymal lining and interstitial spaces and adhering to leptomeninges at near and distant sites. 9,10,42 A study conducted by Tabori et al. identi ed an increased rate of epidermal growth factor receptor (EGFR) ampli cation in dPLGGs compared to non-disseminated PLGGs which seem to suggest the possible role of EGFR ampli cation in tumor dissemination. 43 The increased EGFR ampli cation promoting tumor growth and the invasive potential of tumor cells may potentially promote tumor dissemination. 44,45 In other CNS tumor types, some biomarkers have been identi ed to promote tumor metastasis. Overexpression of ERBB2 (also known as HER2) leading to activation of the PI3K/AKT signaling pathway and ERK1/2 pathway which results in the up-regulation of S100A4, a pro-metastatic gene have been found in metastatic medulloblastoma compared with localized medulloblastoma. 46,47 S100A4 has also been found to be up-regulated more often in ependymoma and glioblastoma than in low grade astrocytoma. 48, 49 It is thus possible that ERBB2 may act in synergy with genetic alterations affecting the RAS/MAPK signaling pathway while also promoting PLGG dissemination via the PI3K/AKT and ERK1/2 pathways. Increased expression of PDGFR, a tyrosine kinase has also been found in low grade gliomas. 50,51 PDGFR promotes glioma stem cell migration and invasion through increasing MMP-2 activity. 52,53 PDGFR-induced MMP-2 activity in low grade gliomas may contribute to tumor dissemination. While understudied and largely speculative within PLGGs, the drivers of metastasis are known for other CNS tumors. Their possible connection with PLGG dissemination is an intriguing prospect that warrants attention.
Treatment approach for dPLGGs was observed to be similar to that of non-disseminated PLGGs for all treatment modalities. Neurosurgical intervention was largely in uenced by the timing of dissemination and the location of primary lesion if identi ed. The majority (72%) of cases with primary disseminated tumors only had a biopsy done and the remaining 28% had either a gross total resection (GTR) or subtotal resection (STR) of a dominant lesion. As expected, the reverse was observed for cases with secondary disseminated tumors with the majority (65%) of cases having either a GTR or STR and 35% undergoing biopsy alone. A less invasive surgical approach is typically favored in patients with primary dissemination wherein the primary goal is to obtain diagnostic tissue rather than attempt curative excision of lesions. An exception is when a dominant lesion is causing symptoms related to mass effect, edema, or cortical irritation, and surgical resection or debulking facilitates symptom resolution and/or adjuvant therapy initiation. Identifying tumors with dissemination potential would help predict prognosis more accurately and allow improved counselling for patient.
Adjuvant therapy is indicated in almost all cases of dPLGG. The predominant chemotherapeutic agents for rst line therapy were vincristine and carboplatin.
This treatment combination has been shown to be effective in the management of PLGG. 5,54,55,56 Second line treatment in cases of disease progression or adverse drug reactions was far more heterogeneous (Table 4). Few children (1.4%) received targeted therapies such as BRAF or MEK 1/2 inhibitors as a second line regimen based on reports from recent studies. Radiotherapy was utilized less frequently, as second or third line therapy for disease progression despite multimodal chemotherapy. The few cases who received radiation as rst line therapy were all older children with ages ranging between 3 and 16 years.
The clinical course of dPLGGs tends to be protracted and may require multiple interventions including salvage therapy for disease progression. Compared to non-disseminated PLGG, dPLGG is associated with a worse outcome. 5 There are several limitations to be considered in the interpretation of results presented in this review. The clinical use of variable terminologies to describe this tumor type may have in uenced our search results. Beyond including the most common descriptors, we address this by reviewing the reference lists of articles initially identi ed to nd additional articles which may not have been captured by the original search terms. There was a lack of comprehensive reporting on the genomic pro le of tumors to draw meaningful conclusions on the pattern of molecular alterations found in all dPLGGs. Finally, many of the included studies were case reports and small case series that carried a measurable risk of bias as determined by the ROBINS-I tool. Based on our GRADE assessment of quality of evidence, all studies reviewed either ranked low or moderate in quality with none ranking high. This underscores the need for clinical trials to speci cally include and sub-analyze dPLGG cohorts to better understand the e cacy of PLGG regimen and targeted molecular therapies in the management of this disease.

Conclusion
While 1p deletion and BRAF-KIAA1549 fusion in dPLGGs have been commonly reported in the literature, these tumors appear to express heterogeneous molecular and biological characteristics distinct from non-disseminated PLGGs. This review suggests that the presence of disseminated disease may not necessarily confer a poor prognosis for all patients as previously noted in earlier reports. There is however lack of comprehensive and quality studies characterizing the molecular makeup of dPLGGs and how treatment approaches including the use of targeted therapy impact survival outcome. Additional studies on the molecular and biological features of these tumors are needed to better understand the pathogenesis of dissemination of PLGG and inform the development of additional targeted regimens to further improve outcomes.