A 22-year-old male referred to the emergency department in July 2006 for fluctuating paresthesia with motor dysfunction of the left arm and leg associated with cervical pain, which had been evolving for 1 year. Otherwise, his medical clinical history was unremarkable. Magnetic resonance imaging (MRI) of the spine revealed a suspicious lesion within the left spinal cord at the levels of C3–C5. The patient underwent subtotal resection in August 2006. An MRI of the spine 1 month following surgery showed a residual tumor of 27 × 8 mm with strong patchy enhancement following gadolinium administration within the left spinal cord at the level of the bottom of C3 to the top of C5, isointense T1 signal, and heterogeneously hyperintense T2 signal. There was an associated syringomyelia at the rostral and caudal aspects of the enhancing tumor, mostly from C2 to C7. Moreover, T2 hyperintensity was observed in the spinal cord above and below the syringomyelia without associated enhancement (Figure 1).

The histological examination of the lesion showed a tissue with mixed glial and neuronal components (Figures 2A,B), the presence of fusiform cells with anisonucleosis, sustained by blood vessels with thickened wall surrounded by lymphocytic cuffs, with eosinophilic granular bodies, and Rosenthal fibers. Binucleated neurons were visualized by calretinine, neurofilament, and synaptophysin staining. Many glial cells showed S100 and CD34 immunoreactivity and diffuse glial fibrillary acidic protein. The Ki-67 labeling index was very low (<1%) and some parts of the tissue were positive for P53 in immunohistochemistry analysis. Molecular analysis revealed immunoreactivity to isocitrate deshydrogenase gene 1 (IDH1 R132H) and a loss of chromosome 9p. Despite the presence of an IDH mutation, central pathological review led to the diagnosis of WHO grade I ganglioglioma (1).

Postoperatively, the patient maintained his neurological symptoms and had Brown-Sequard syndrome and micturition dysfunctions.

The patient was followed up with for 9 years until MRI demonstrated tumor progression. Tumor measurements were then 36 × 12 mm, corresponding to a 50% increase in size (Figure 1). At this time, a second resection was considered too risky and gross tumor resection was not possible. No other treatment was administered because of the lack of proof of chemotherapy and radiotherapy interest. This was consistent with increased arm and leg dysfunction.

Molecular testing for evaluation of target therapy was implemented using tissue collected during surgery after obtaining written informed and signed consent. In July 2015, genomic DNA was extracted from the tumor tissue with a QIAamp® DNA mini Kit (QIAGEN, Hilden, Germany) for standard direct sequencing of exon 15 of BRAF, which was analyzed by using a SNaPshot® kit (Thermo Fisher Scientific, Waltham, MA, USA). The results revealed a V600E BRAF mutation and no mutation in RAS.

Based on these results, in November 2015, the patient was started on vemurafenib 960 mg orally twice daily [100% of the recommended dose in adults for melanoma (24)]. This treatment was determined as part of “AcSé,” a French program known as “Secure access to innovative targeted therapies” (38). After 8 weeks of treatment, the patient was neurologically stable and brain MRI showed a >50% decrease in tumor size (Figure 1). A steady partial response was observed for more than 13 months. Toxicities were measured by the Common Terminology Criteria for Adverse Events v4.0 and included grade I myalgia, arthralgia, and asthenia as well as grade I maculopapular rash (folliculitis with microcysts on legs and arms treated with topical retinoids). After 13 months of treatment, the patient decided to stop the treatment because of grade II photosensibility and other dermatological side effects (Figures 3A–C). To manage his rashes, folliculitis, and microcysts, the patient applied glycerol as a topical emollient, 30% pure urea cream, and Trétinoine (topical retinoid). His palmar-plantar erythrodysesthesia syndrome (hyperkeratosis) was treated with topical fluorouracil/salicylic acid and even curettage for some areas. Photoprotection was achieved by applying sun cream during treatment. No topical steroid was used. His Eastern Cooperative Oncology Group Performance Status decreased to 2 because of grade II asthenia. Two months after stopping treatment, MRI revealed that the disease was stable and had not significantly progressed according to RANO criteria (39) (Figure 1). Six months after stopping vemurafenib, grade I dermatological side effects persisted but the patient had recovered to a normal Performance Status and MRI showed no signs of progression.

Twenty-one months after vemurafenib discontinuation in October 2018, MRI and neurological examination showed stable disease (Figure 1) and the patient had no side effects. Although the disease is incurable nature, his neurologic and cognitive functions and quality of life were preserved.

The first description of ganglioglioma was detected in 1870 by Loretz and further studied in 1926 by Perkins. Ganglioglioma is a rare tumor of the central nervous system accounting for 1–1.5% of all spinal tumors (4, 40, 41). Gross total resection is the most reliable treatment (10, 42). While the larger part of this disease occurs in the temporal lobe, causing epilepsy (5) and showing a male preference, its spinal location makes treatment difficult, increasing the risk of recurrence (10, 42). Dudley et al. used the large Surveillance Epidemiology and End Results database, which represents nearly one-third of North America's population, and identified 348 children with low-grade gangliogliomas to study their characteristics (8). This was the largest study to evaluate the spinal location of this rare tumor. Spinal cord gangliogliomas represented 3.5% of cases, with nearly 100% of survival at 5 years and the highest percentage of gross total resection of more than 83%.

MRI findings for supratentorial ganglioglioma can be divided into three groups: cystic, cystic-solid and solid (43). For intramedullary ganglioglioma, imaging manifestation varied considerably (44). In a recent study of 142 cases, all gangliogliomas in the cervicomedullary junction and all BRAF mutation-positive ganglioglioma were contrast-positive (21). Our case had a solid lesion with patchy enhancement and a cystic component, which was consistent with previous reports (45). Furthermore, the rapid but not significant regrowth of the tumor after treatment discontinuation in our case may be associated with a “rebound effect,” as described previously (30, 37). This is analogous to pseudo-progression. Indeed, pseudo-progression is commonly observed in asymptomatic patients and occurs at weeks and up to 3 months after treatment. However, previous studies showed that pseudo-progression occurs because of radiotherapy and is characterized by transient T1 gadolinium enhancement resulting from breakdown of the blood brain barrier, which typically resolves spontaneously without treatment (46). Pseudo-progression has also been described in patients treated with immunotherapy, but its incidence is unknown because of the lack of available data. In the two previous reports (30, 37), the therapeutic benefit was again achieved after vemurafenib re-introduction. Re-activation of the Ras/Raf/MEK/ERK/MAP kinase pathway may occur, but the biological mechanism remains unclear.

In a retrospective review of 58 patients (median age at diagnosis of 8.5 years) who underwent surgical resection, the 5- and 10-year overall survival rates were 89 and 83%, respectively. The spinal cord location was associated with a 3.5-fold increased risk of recurrence compared to cerebral gangliogliomas (47).

The efficacy of chemotherapy for adjuvant or recurrent ganglioglioma is uncertain and remains controversial (48), with a high risk of serious adverse events. Recommendations for the use of radiotherapy at progression are based on case reports and small cohorts, particularly in the spinal cord (49). Radiotherapy may result in a better local control for subtotal resection in the supratentorial location, but does not improve overall survival (4, 5, 11–15). Some case reports even suggested that radiotherapy can result in malignant transformation (16, 17). Based on these reports, we did not treat our patient with radiotherapy or chemotherapy.

The presence of the BRAF V600E mutation suggests that use of BRAF inhibitors are efficient for treating recurrent gangliogliomas. BRAF mutation appears in 8% of human cancers (50). The BRAF V600E mutation was found more often in pediatric low-grade than in high-grade gliomas (2, 18, 51), likely because low-grade gliomas are the most frequent brain tumors in children (52). Patients with BRAF V600E mutation exhibit shorter progression-free survival (22, 53). However, the prognostic value of this mutation in recent studies is controversial (54). Thereby, Jones et al. suggested that caution should be used when interpreting the BRAF mutations status as an independent prognostic marker (55).

BRAF V600E mutations were detected in nearly 20% of gangliogliomas in a screen of 1,320 nervous system tumors (19). This was the second most frequently BRAF-mutated cerebral tumor entity after pleomorphic xanthoastrocytoma. In another cohort, 50% gangliogliomas were mutated (20). In a recent series, BRAF V600E mutations were detected in 38% of cases but all spinal cord gangliogliomas were wild-type (56). Another group identified only two tumors among 19 (10%) intramedullary gangliogliomas harboring a BRAF V600E mutation (3).

Young adult age, synaptophysin positive tumor, lymphocytic cuffs, and a high Ki67 level (mean 2.5%) have been shown to be associated with the BRAF V600E mutated status (57). However, the results of Ki67 analysis did not reach statistical significance. Moreover, the mutation appears to be present in the neuronal component or both the neuronal and glial components, but never in the glial component alone (5, 57–59). IDH mutations were reported in 8% of cases in a series of 100 gangliogliomas (60). The presence of this mutation was correlated with a greater risk of recurrence and malignant transformation. In the 2016 WHO classification, detection of IDH1 mutation in a tumor resembling a ganglioglioma strongly supported the diagnosis of an infiltrating glioma with ensnared neurons (1, 61). However, as observed in our patient, it has been increasingly recognized that some circumscribed gliomas can harbor mutations typically encountered in diffuse gliomas (such as IDH and histone mutations) (62–65). Occasionally, H3K27M mutations have been reported in midline gangliogliomas (66). In contrast to diffuse gliomas, H3K27M mutations do not appear to be associated with a poor prognosis in circumscribed gliomas. The H3K27M mutation status was not determined in our case.

There are some previous descriptions of the efficacy of vemurafenib and dabrafenib (another BRAF inhibitor) in low- and high-grade gliomas other than ganglioglioma (35, 67–71). In a basket study with vemurafenib in BRAF V600E mutation-positive non-melanoma cancers (35), the objective response rate in BRAF-mutant gliomas was 25%. Previously reported cases of gangliogliomas treated with a BRAF inhibitor are listed in Table 1 (26–37). The response to vemurafenib in our patient was consistent with the response to BRAF inhibitors observed in previously reported cases, including one case of spinal ganglioglioma in a 2-year-old child (29). However, all cases except for one were located in the cerebrum or were brainstem gangliogliomas, with half of the cases being anaplastic gangliogliomas (9/19) (28, 32–36). In eight cases (8/19) (26, 27, 32–34, 36, 37), the BRAF inhibitor was associated with another treatment or surgery, making the analysis of the response to the BRAF inhibitor difficult. Based on the analysis of the present case and previously reported cases, a complete response was obtained in 15% (3/20) and partial response in 50% (10/20) of cases at a median of 3.2 months after starting treatment and the estimated progression-free survival was 14 months. In 12 patients in whom a BRAF inhibitor was administered as a single agent, the response rate was 50% (6/12) (one complete response and partial response in all other patients). Additionally, 33% (4/12) showed stable disease and 17% (2/12) showed progressive disease. The estimated progression-free survival was 11 months. The median follow-up time after starting treatment was 14.5 months, while this time period was 36 months in our case, including 21 months of stable disease after discontinuation. The present case is remarkable because our patient had spinal ganglioglioma treated with vemurafenib alone and a long follow-up. Interestingly, 2 months after vemurafenib disruption for patient convenience, a moderate (<25%) increase in the size of the contrast enhancement was observed, after which the tumor remained stable in subsequent MRIs. This rapid but not significant regrowth was consistent with the previous report of a “rebound effect” following vemurafenib disruption after protracted exposure to this treatment (30). In this situation, vemurafenib re-challenge may be effective (30), but the present case suggests that close follow-up is another option, as further tumor progression may not systematically occur. In recent years, BRAF/MEK double blockade with vemurafenib and cobimetinib or dabrafenib and trametinib was shown to be a more effective strategy than targeting BRAF alone in patients with BRAF-mutant advanced melanoma (72). Dual BRAF/MEK inhibition has also been suggested as a promising activity in BRAF-mutant gliomas that may overcome (36, 73) vemurafenib resistance. A prospective study is needed to assess the efficacy of this combination in gangliogliomas.

In our case, the patient asked for treatment discontinuation because of dermatological toxicity. To avoid treatment discontinuation, intermittent dosing could have been used, which has been shown to result in persistent efficacy and improve tolerability as a means of managing BRAF inhibitor toxicity (74). Another possibility may have been switching the patient to another, better-tolerated BRAF inhibitor such as dabrafenib or combining the treatment with an MEK inhibitor which is paradoxically associated with fewer secondary effects than BRAF inhibitors alone (33).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.