Fibroblast growth factor receptor 1 gene mutation as a potential risk factor for spontaneous intracranial hemorrhage in pediatric low-grade glioma patients

Abstract Background Fibroblast growth factor receptor 1 (FGFR1) mutations have been associated with poorer prognoses in pediatric central nervous system tumor patients. A recent study highlighted a link between FGFR1 mutations and spontaneous intracranial hemorrhage (ICH), demonstrating that all patients with an FGFR1 alteration experienced hemorrhage at some point during their course of treatment. Methods The current study examined 50 out of 67 pediatric patients with low-grade gliomas (LGGs) who had genomic testing between 2011 and 2022 at our institution to determine whether a correlation exists between FGFR1 mutations and spontaneous ICH. Results We found that of the 50 patients with genomic data, 7 (14%) experienced ICH, and an additional spontaneous hemorrhage was recorded; however, no genomic testing was performed for this case. Five of the seven patients (71.4%) had an FGFR1 modification. In our patient population, 6 expressed a detectable FGFR1 mutation (66.7% [4/6] had N546K alteration, 16.7% [1/6] FGFR1 exons duplication, and 16.7% [1/6] had a variant of unknown significance [VUS]). The patient with the FGFR1 VUS had no reported spontaneous hemorrhage. Statistical analysis found a significant association between FGFR1 and spontaneous intracranial hemorrhage (P-value = < .0001). In the patient population, all cases of PTPN11 alterations (n = 3) co-occurred with FGFR1 mutations. Conclusions Our case series highlights this link between the FGFR1 mutation and spontaneous intracranial hemorrhage in pediatric LGGs.

Tumors of the central nervous system (CNS) are the second most common cancer in children and remain the leading cause of cancer-associated mortality in pediatric populations. 1 Of the diverse types of pediatric CNS tumors, low-grade gliomas (LGGs) are the most common and comprise nearly one-third of all cases. 2 Overall 5-year survival (OS) for children with LGGs is excellent even with incomplete tumor resection. 3However, these children require treatment with chemotherapies, radiation, and/or additional surgeries all of which are associated with morbidities that significantly impact the quality of life for patients, including long-term effects on the endocrine system and neurodevelopment. 4ediatric LGGs are associated with genetic alterations in the mitogen-activated protein kinase (MAPK) signaling pathway. 3Most Fibroblast growth factor receptor 1 gene mutation as a potential risk factor for spontaneous intracranial hemorrhage in pediatric low-grade glioma patients frequently, they harbor mutations within the pathway at the Braf proto-oncogene (BRAF) gene, specifically the BRAF V600E missense mutation and the KIAA1549:BRAF fusion. 4The second most common class of genetic alterations in pediatric LGGs occurs at the fibroblast growth factor receptor 1 (FGFR1) gene, believed to drive the proliferation of multipotent stem cells in developing human brains. 2,5FGFR1 alterations, including the FGFR1 N546K and K656E missense mutations, constitutively activate this pathway. 6In several pediatric CNS tumor types, including pilocytic astrocytoma (PA)-the most common type of low-grade glioma, FGFR1 mutations have been linked to poorer prognoses. 7 recent study suggested an association between FGFR1 mutations and spontaneous intracranial hemorrhage (ICH) in pediatric LGG patients. 8Although rare, intracranial hemorrhage can have devastating effects on pediatric populations.An estimated one-third of pediatric patients who experienced an ICH die, 9 and those that survive may suffer neurodevelopmental effects 10 from long-standing neurologic, cognitive, and adaptive behavior impairments affecting their potential independence and social functioning. 11In the aforementioned study, Ishi et al. found that all reported patients (n = 4) with an FGFR1 mutation experienced spontaneous ICH at some point during their course of treatment. 8The significance of this finding could be highly impactful, as tumor-specific risk factors for spontaneous intracranial hemorrhage have yet to be identified. 8However, more studies with additional patients harboring FGFR1 mutations are needed to further confirm these findings.
In the present study, we aim to contribute by presenting data from 50 patients treated for LGGs at our institution during 2011-2022 that had genomic testing performed.We investigate the frequency of spontaneous intracranial hemorrhage and link these events to their genetic mutations, specifically looking for any statistically significant correlation between hemorrhage and FGFR1 mutations.Furthermore, we present any cases in which a patient with an FGFR1 mutation experienced spontaneous intracranial hemorrhage, with the hopes of contributing to the current literature on this potential novel risk factor in pediatric LGG patients.

Patient Population
For this retrospective study, we included 67 pediatric cases from patients diagnosed and treated for LGGs at the Orlando Health Arnold Palmer Hospital for children from 2011 to 2022.The selection of the patients depended on their age (≤18 years) and the diagnostic grade for LGGs (grades I and II), previously used by the World Health Organization (WHO) to classify malignancies.Patient information accessed for this study included treatment courses, clinical outcomes, and radiographic and pathological findings.A secondary filtering was performed for patients who underwent genetic and molecular analyses at the time of tumor resection and biopsy.Fifty patients who fit the criteria remained and each had one or a combination of the following molecular analyses: Neuro-Oncology Expanded Gene Panel, Chromosomal Microarray, Methylation, and Next-Generation Sequencing.All patients examined in this study were consented under the IRB-approved Translational Research Protocol.

Statistical Analysis
The statistical Fisher's exact 2-sided test followed by the Baptista-Pike odds ratio method was performed with GraphPad Prism version 9.4.1 (GraphPad Software).A P-value < .05 was considered statistically significant.

Genetic Mutations and Spontaneous Hemorrhage in Low-Grade Gliomas
Of the 50 patients with pediatric LGGs, only 6 cases have an FGFR1 mutation (12%).However, 2 of the 6 patients do not exhibit the typical FGFR1 point mutations, N546K or K656E, but present with either a duplication of exons 9-18 or a variant of unknown significance (VUS), R207H.The patient with a VUS also had a KIAA1549:BRAF fusion and did not present with an intracranial hemorrhagic stroke (case not discussed).Three out of fifty patients (6%) have PTPN11 mutations (N58D, E69K, and E76K), all co-occurring with FGFR1.Twenty-seven of the fifty (54%) patients have a BRAF gene alteration (28% have the KIAA1549 fusion, 20% have the BRAF V600E mutation, 4% have the PRKARB2B fusion, and 2% present with a BRAF duplication).Seven patients (14%) underwent molecular analysis, but no pathogenic mutations were detected.The remaining 11 patients (22%) had other mutations not discussed in this case study (BRCA1, ERCA2-RAF1, IDH1, KRAS, MEK2, PIK3R1, RAF1, SDHC, SETBP1, and TERT).

Importance of the Study
For pediatric low-grade gliomas (pLGGs), FGFR1 gene alterations have been linked to poorer prognosis and have been correlated with the manifestation of intracranial hemorrhage (ICH) in one reported cohort of patients.The study reports a statistically significant relationship between spontaneous hemorrhage in pediatric patients with FGFR1-altered LGG.The genomic profiles show a concurrent expression of PTPN11 alteration with FGFR1, where all the cases exhibited ICH.The following clinical case series strengthens the link between FGFR1 alterations and spontaneous ICH.

Gonzalez-Vega et al.: FGFR1 gene mutation as a potential risk factor for spontaneous ICH in pLGGs patients
6 patients with an FGFR1 mutation manifested a spontaneous hemorrhage at some point during their course of treatment.The remaining 2 patients with spontaneous hemorrhage expressed either a BRAF V600E mutation or a PRKAR2B:BRAF fusion.Due to the frequency and interest of the FGFR1 mutation in spontaneous hemorrhage cases, we present cases 1 through 5 where we observed the co-occurrence of the events (Table 1).

Case 1
The male patient originally presented at age 9 with decreased vision and a history of Noonan's Syndrome.Initial MRI imaging detected an optic pathway glioma, and the patient was started immediately on Carboplatin and Vincristine as per the Children's Cancer Group (CCG)-A9952 regime A. He transferred to our institution the following year and continued treatment until MRI showed disease progression.He underwent a biopsy confirming the diagnosis as a PA and started a weekly Vinblastine treatment.Three months later, MRI again showed tumor progression and he underwent tumor debulking and cyst fenestration.He was started on Avastin and Everolimus for 12 cycles and continued Avastin due to tumor response.However, treatment was stopped due to intratumoral hemorrhage.He then received 14 cycles of Everolimus and 10 cycles of Carboplatin.Six months routine MRI again showed progression and the patient was treated with Trametinib, which he tolerated for 11 months but was then stopped due to a shunt revision.
At age 16, the patient underwent a second sub-total resection and molecular sequencing with reported positive mutations for FGFR1, MEK2, PTPN11, and NF1 genes.The patient was diagnosed with progressive disease 3 months after surgery and treated under the NCI-Children's Oncology Group (COG) MATCH Study with Erdafitinib, an FGFR1 inhibitor, which initiated tumor response but was not tolerated by the patient.Due to his previous response, Avastin was restarted, but he again developed a large intracranial hemorrhage that made his tumor appear 3 times the size (Figure 1).He underwent a third partial resection followed by partial radiation, which was stopped due to another small hemorrhage.A decision was made to stop treatment, and the patient succumbed to his disease at the age of 18.

Case 2
A female patient was diagnosed at age 5 with an optic pathway glioma.She underwent a biopsy and pathology analysis was consistent with a PA.She began treatment with Vincristine, Carboplatin, and Temozolomide as per Children's Oncology Group (COG)-ACNS0223.MRI showed progressive disease 2 years later and she was started on monthly Temodar for a year.The following year, MRI once again showed progression and she was started on Avastin and Irinotecan.Once again, a year later progression was detected, and the patient underwent MRI-guided laser ablation of the tumor.After several months, she came to the ER unconscious and exhibited intracranial hemorrhage and cardiac arrest.At age 14, she transferred to our institution, where her tumor showed progressive disease and had a 40% increase in size.Her tumor tissue was sent for molecular testing and 2 mutations were detected in the FGFR1 and PTPN11 genes.Based on these findings the patient was started on Everolimus for 12 cycles, but once again the patient showed progressive disease.She was then started on Trametinib but switched to Vinblastine due to toxicity.She recently finished her treatment with Vinblastine and currently has stable disease.

Case 3
The male patient originally presented at age 8 with precocious puberty.MRI found a pituitary lesion that was thought to be a microadenoma, which was followed without changes.Three years later the patient had a sudden onset of headaches and emesis.MRI found a hemorrhagic suprasellar mass (Figure 2).He was referred to our institution, where he underwent a partial resection of the tumor.Pathology was consistent with a low-grade glioma, which was confirmed through molecular testing.Nextgeneration sequencing found that the tumor harbors an FGFR1 and a RAS mutation.He was started on Carboplatin and Vincristine as per CCG-A9952A, and he is currently on cycle 8 with stable disease.

Case 4
Male patient presented at age 14 with acute headaches and an altered mental status.CT scan showed a suprasellar mass with intratumoral hemorrhage, as well as intraventricular hemorrhage (Figure 3).He underwent a biopsy, and the pathology was consistent with PA with FGFR1 (p.N546K) and PTPN11 mutations.MRI revealed leptomeningeal disease.He was treated with Carboplatin and Vinblastine as per protocol COG-ADVL0515.After 4 months of treatment, he had progressive disease and was started on Carboplatin, Vincristine, and Temozolomide as per COG-ACNS0223.Due to behavioral concerns, therapy was changed to CCG-A9952, omitting Vincristine due to neuropathy.The patient completed 8 cycles of therapy and had stable disease for 20 months.He then progressed and was treated with Everolimus for 9 months until the MRI again showed progressive disease.He was started on singleagent Temozolomide (200 mg/m 2 days 1-5 every 28 days), but progressive leptomeningeal disease (LMD) was observed 6 months later.The patient is currently on therapy with Carboplatin and Temozolomide as per COG-ACNS0223.

Case 5
Female patient presented at almost 1.5 years old with an acute midbrain hemorrhage that was initially thought Gonzalez-Vega et al.: FGFR1 gene mutation as a potential risk factor for spontaneous ICH in pLGGs patients to be a cavernoma (Figure 4).After the hemorrhage resolved, she was diagnosed with a brain tumor.The patient underwent 2 biopsies, though both were insufficient for diagnosis.That same year, she transferred to our institution after declining neurologically and developing hydrocephalus, where she underwent tumor debulking.
Pathology was consistent with PA of the midbrain/ pons and cerebellum, and molecular sequencing of the tumor revealed an FGFR1 mutation.The patient was initially treated with Carboplatin and Vinblastine but was switched after 1 cycle to a weekly Carboplatin regimen per CCG-A9952 regimen A (without Vincristine) and completed 8 cycles.After 19 months, she was found to have progressive disease, and treatment was restarted as per CCG-A9952 Carboplatin with desensitization protocol; however, further doses were held due to safety concerns.She was started on weekly Vinblastine for 2-12-week cycles, then every 2 weeks for 2-12-week cycles for 1 year of treatment.
Avastin was added later every other week.Several months later, MRI again revealed progression, and treatment was changed to oral Trametinib, which she took for 7 months until progression was again noted.Her treatment was then changed to oral Temodar.Three months later she again progressed, and her treatment was changed to Trametinib and Everolimus.

Pediatric LGGs Presenting Spontaneous Hemorrhage and the Factors Involved
The patient population has a stable age distribution for diagnosis and clinical presentation.Additionally, spontaneous hemorrhage in the study's population does not favor a specific tumor location such as the optic pathway, brain stem, temporal lobe, parietal lobe, suprasellar region, and diencephalic region (P-value = .089,.370,> .999,.370,.263,and .140,respectively).Additionally, spontaneous Gonzalez-Vega et al.: FGFR1 gene mutation as a potential risk factor for spontaneous ICH in pLGGs patients hemorrhage shows no correlation with tumor pathology in the patient population of this study.Although the highest occurring mutation reported in this population involves BRAF alterations (27/50, 54%), there is no correlation between a disease-provoked intratumoral hemorrhage and BRAF.In contrast, both FGFR1 and PTPN11 gene alterations have correlated with spontaneous hemorrhage (P-value= < .0001,and .0018;Table 2).Interestingly, the PTPN11 mutation exclusively occurred as a co-mutation with FGFR1 in all the reported cases in this study.

Discussion
To date, tumor-specific risk factors for spontaneous intracranial hemorrhage in pediatric LGGs have yet to be identified.Ishi et al., (2020) proposed a link between the FGFR1 mutation and spontaneous hemorrhage events after they determined that all their patients with an FGFR1 mutation (n = 4) experienced hemorrhage at some point during their course of treatment.FGFR1 mutations are the second most altered gene in pLGGs and occur in 5%-10% of pLGG cases. 2 In our patient population, 6 exhibited a detectable FGFR1 alteration (12%).3][14] Only 2 of the 6 FGFR1-altered patients did not have any of the common alterations and instead expressed a duplication associated with increased activation of the pathway 15 or a VUS that has yet to be regarded as pathogenic.The one patient with the VUS was the only FGFR1 mutant with no reported spontaneous hemorrhage.Ishi et al., (2020)  found a statistically significant correlation between the FGFR1 alterations and these spontaneous intracranial hemorrhagic (ICH) events.Their findings favor the FGFR1 K656E mutation, 3 out of the 4 patients (75%), and 1 of the patients (25%) presented with the N546K alteration.
Correspondingly, the present study found that 4 out of the 5 cases (80%) with both ICH and FGFR1 mutants had the N546K alteration, with the extra case not being a point mutation but exons 9-18 duplication.Additionally, a statistical significance was established between intracranial hemorrhage and the PTPN11 gene, which often co-occurs with FGFR1, suggesting a cooperative role of these mutations. 16nlike Ishi et al., (2020), who found a statistical significance between hemorrhagic events and the diencephalic region, we did not see any correlations between intracranial hemorrhage and the location of the tumor.
Although there is currently no biological explanation for the association between FGFR1 mutations and spontaneous hemorrhage, the cases highlighted by Ishi et al., our current study, and other case studies 17 illustrating this link warrant the need for future investigation into the potential mechanisms behind it.Both FGFR1 and FGFR2 have been reported to be critical embryonic developmental factors, crucial for survival and injury response. 18Defects in the FGF/FGFR signaling pathway have been closely correlated with cardiovascular diseases due to the modulation of angiogenic and neovascularization responses. 19Furthermore, a tumor cancer manifestation in the CNS can be cataloged as an injury, meaning a vascular response occurs in many of the cases and an impaired FGF/FGFR signaling pathway such as a point mutation on the gene will affect the maintenance of neovascular vessels.Additional case studies with a larger number of patients are needed to directly link FGFR1 mutations and spontaneous hemorrhage in pLGGs.

Conclusion
Although uncommon, spontaneous intracranial hemorrhage can have devastating effects on pediatric low-grade glioma patients.To date, little is known about the tumorrelated risk factors for these hemorrhagic events.However, a recent study proposing the FGFR1 mutation as a potential novel risk factor for spontaneous intracranial hemorrhage in pLGG patients warrants further studies.The current study is consistent with their results, giving weight to their findings and validating the need for additional investigation to discover the specific mechanisms behind the association.This knowledge could help clinicians better predict the occurrence of spontaneous hemorrhages and provide further insight into the biology of these tumors to lead us toward continued improvement in our targeted treatments.

Figure 1 .Figure 2 .
Figure 1.Case 1 brain CT scans without contrast show an intratumoral hemorrhage in the third ventricle.(A) Axial view of the brain with an arrow indicating the intratumoral hemorrhage.(B) Sagittal view of the brain with arrows indicating a shunt port (top) and the intratumoral hemorrhage with enlargement of the third ventricle (bottom).(C) Hematoxylin and eosin (H&E) stained slide image depicting a low-grade spindle cell neoplasm with alternating hypercellular and hypocellular areas (characteristic of a Pilocytic Astrocytoma).A, Anterior; S, Superior; R, Right.Scale bar = 50 µm.

Figure 3 .Figure 4 .
Figure 3. Case 4 brain CT scans without contrast showing an intratumoral hemorrhage.(A) Axial view of the brain with arrows indicating the intratumoral hemorrhage.(B) Sagittal view of the manifested intratumoral hemorrhage (arrow).(C) Hematoxylin and eosin (H&E) stained slide image at low magnification depicting a low-grade spindle cell neoplasm with alternating hypercellular and hypocellular areas (characteristic of a pilocytic astrocytoma).A, Anterior; S, Superior; R, Right.Scale bar = 50 µm.

Table 1 .
Summary of Pediatric LGG Patients Presenting Spontaneous Hemorrhage PA, pilocytic astrocytoma; *patient presents with variants of uncertain significance, not reported here.

Table 2 .
Univariate Analysis of Associations With Spontaneous Hemorrhage of the Patient Population With Genomic Data Using Fisher's Exact Test PA, pilocytic astrocytoma; PMA, pilomyxoid astrocytoma; Inf, infinite; P-B Fusion, PRKAR2B:BRAF Fusion.Italics type P-value indicates statistical significance (P < .05).