BRAF V600E mutation in Juvenile Xanthogranuloma family neoplasms of the central nervous system (CNS-JXG): a revised diagnostic algorithm to include pediatric Erdheim-Chester disease

In the most recent revised classification of histiocytic disorders, [21], cutaneous juvenile xanthogranuloma (JXG) lesions and those JXG lesions with a systemic component, but not associated with a molecular alteration, are categorized separately into the cutaneous or “C”-group histiocytosis. However, extracutaneous JXG lesions with mitogen activated pathway kinase (MAPK) / extracellular-signal-regulated kinase (ERK) pathway activating mutations are now categorized into the Langerhans “L-group” histiocytosis, including three rare BRAF V600E JXG “L-group” neoplasm [56]. In this revised classification, Langerhans cell histiocytosis (LCH) and Erdheim Chester Disease (ECD), are also categorized in the “L group” of histiocytic neoplasms. At the far ends of their phenotypic spectra, LCH, ECD, and JXG all have distinct clinical and pathologic features; however, this shared categorization was proposed based on similar molecular alterations, mixed LCH/ECD histiocytic presentations in adult cases, and accumulating data supporting a common hematopoietic precursor, at least between adult LCH and ECD [21]. However, pediatric extracutaneous JXG with MAPK molecular alterations as an L-group histiocytosis, has been less studied in relation to its possible shared origins with LCH and pediatric ECD [10, 16, 38, 40, 46, 51] Furthermore, while the BRAF V600E mutation constitutes the majority of molecular alterations in ECD and LCH [3, 5, 30, 53], only three reported JXG neoplasms, all in male pediatric patients with localized central nervous system (CNS) involvement, are known to harbor the BRAF mutation; however, none showed evidence of systemic disease or a prior history of LCH [56].

In general, CNS-JXG neoplasm are rare, often requiring surgical resection or chemotherapy [13, 36, 55, 58] and do not have the propensity to regress spontaneously, unlike their cutaneous JXG counterpart [58]. CNS-JXG neoplasm range from isolated CNS lesions to multifocal CNS lesions to those associated with systemic disease [6, 13, 22, 26, 27, 36, 58]. In adults, CNS based neoplasms with a JXG or xanthogranuloma pathologic phenotype are often the first and most debilitating manifestation of ECD. They are often a challenge to diagnose and have a generally poor prognosis; however, in adults these neoplasms often have an excellent response to inhibitor therapy [15, 24, 48]. In children, both systemic JXG with CNS involvement and CNS-limited JXG also appear to have poorer outcomes, as compared to pediatric JXG without CNS disease; however, none of these prior pediatric JXG studies have investigated the BRAF mutational status [13, 58].

Furthermore, the current revised classification of histiocytes [21] has created a divide between the JXG family of neoplasms with molecular alterations (L-group) and those without molecular alterations (C-group). Standing alone, this grouping does not have particular clinical significance, especially given that both C-group and even R-group histiocytic lesions now also harbor MAPK-pathway activated mutations [16, 25, 28, 44, 49, 52]. Furthermore, the World Health Organization (WHO) recommends that CNS neoplasms have an initial morphologic report followed by an integrated final diagnosis after molecular studies are completed [42]. The aim of this study is to revisit the pathology and incidence of BRAF V600E mutations in pediatric CNS-JXG neoplasms in order to propose a revised diagnostic algorithm that requires the integration of pathology, molecular, clinical, and radiographic findings for a comprehensive final diagnosis, in the hope of advancing clinical management and treatment options.

We identified 22 CNS lesions with JXG phenotype that met our initial inclusion criteria (Fig. 1), designated as CNS-JXG. Fourteen CNS-JXG cases (64%) had informative molecular status for the BRAF V600E point mutation, which included 10 pediatric CNS-JXG neoplasms that were included in the primary analysis. The overall mean age was 7.3 years (range: 1–18 y) with a male/female ratio of 1.5 (Table 2). Within such a small cohort it is difficult to ascertain clinically relevant, statistical difference between the pediatric BRAF V600E (n = 5) and BRAF wild-type (n = 5) CNS-JXG cohorts, but certain trends were noted. There was a similar mean age in both cohorts [BRAF V600E: 7 years (3–12 y) vs BRAF wild-type 7.6 years (1–18 years)]. While the overall male/female ratio in the pediatric CNS-JXG cohort was male predominant (1.5), the BRAF V600E cohort had more males (male/female ratio: 4.0), as compared to the wild type cohort (male/female ratio: 0.67). The pediatric BRAF V600E cohort also had more cases of multifocal CNS disease [BRAF V600E: 3/5 (60%) vs the BRAF wild-type: 1/5 (20%)] along with associated CNS white matter changes and enhancement of nodular lesions (Table 2). The two cases with systemic disease were BRAF V600E positive (Table 2). One had multifocal CNS-JXG disease, including intracranial, sellar, dural, ventricular, and cavernous sinus involvement, along with bilateral long bone sclerosis and confirmatory bone biopsy also with the BRAF V600E mutation (case 3). Thus the integrated final diagnosis with pathology and radiographic correlation was that of pediatric ECD, as previously published [14]. The second case also had systemic disease, with an associated cutaneous BRAF V600E positive JXG lesions. Symmetric CNS white matter changes were present on MRI (Fig. 2i-l), along with an enhancing parenchymal mass; however, there was no evidence of bone involvement or other classic features of ECD. One of the BRAF V600E cases with multifocal CNS lesions had visual decline and panhypopituitarism from sellar/optic chiasm-based masses, while the other had resultant encephalomalacia and brain atrophy with progressive developmental delay and was started on hospice care six years after initial presentation (Table 2). In contrast, the BRAF wild-type cohort had more isolated CNS lesions without mention of associated symmetric white matter changes or reported systemic disease; however, one of these cases did not have long term follow-up after diagnosis (Table 2). The BRAF wild type group also did not have further molecular testing or phosphorylated-ERK staining.

The presence of the BRAF V600E mutation did not appear to confer a selective morphologic pattern (Table 3). Both cohorts displayed varied histologic features within the morphologic (Fig. 2) and immunophenotypic spectrum of JXG-family (Table 3). Nine of the pediatric cases had S100 available for evaluation. Two BRAF V600E cases and three wild-type cases had scattered S100 positive Rosai- Dorfman-Destombes Disease (RDD)-like cells, along with an additional case in wild-type cohort with multinucleated giant cells and rare cells with emperipolesis, despite no S100 expression (Table 3). Half of the pediatric cases had assessment by Ki-67/MIB-1 immunohistochemistry, with an overall low proliferation index (0–15%) when accounting for intermixed inflammatory cells. The two BRAF V600E cases had subjectively lower median proliferation rate (2%), as compared to the three pediatric wild-type cases (15%) (Table 3); however, there are too few cases to draw statistical conclusions on these results. Focal mild cellular pleomorphism was noted in both groups, but there was no evidence of frank anaplasia or diffuse atypia. Only one of the BRAF wild-type cases (case 6) had central ischemic type necrosis (Table 3). The BRAF-VE1 immunostain showed diffuse, strong (2–3+) granular cytoplasmic expression in the majority of the lesional histocytes (> 75%). However, there was variable staining expression noted in the different JXG-histiocyte subtypes, including within a single lesion. For example, diffuse strong (3+) VE1 expression was noted in epithelioid and finely vacuolated JXG cells, diffuse but moderate (2+) expression in foamy/xanthomatous JXG cells, and weak to negative staining (0–1+) in the fibrohistiocytic JXG component that had more heavily xanthomatous/lipidized cells intermixed with fibrosis/gliosis (Fig. 2). All wild-type cases had a mixture of cell types, with negative staining in the epithelioid/finely vacuolated and foamy/xanthomatous JXG cells (Table 3).

Treatment options in pediatric CNS-JXG cases were variable showing a combination of both surgical excision and systemic chemotherapy (Table 2). For most BRAF V600E CNS-JXG cases, the BRAF mutational status was not known at the time of initial diagnosis. Treatments included the following: LCH III based protocol with prednisone/vinblastine for 12 months in unifocal CNS disease of the hypothalamic-pituitary axis, clofarabine and dexamethasone for systemic JXG with multifocal CNS-JXG, anakinra for pediatric ECD that had previously progressed on prednisone/vinblastine for 6 weeks, cladribine for 6 cycles, and clofarabine for 2 cycles [14], and BRAF-inhibitor therapy with dabrafenib for a multifocal CNS disease, which was aggressive and refractory to first line therapy. In this last case, dabrafenib showed an immediate and dramatic clinical response, including complete resolution of hyperventilation and weaning from dexamethasone with interval MRI at 2 months, 4 months and 15 month of therapy, along with a reduction in intracranial size and no new lesions. (Table 2). Case 4 did not have a known BRAF V600E prospectively in his course with progressive CNS white matter disease in the years following excision. The wild-type cases also had surgical resection with initial prednisone/vinblastine and then cladribine in one case with multifocal lesions and prednisone/vinblastine in a unifocal lesion of the cerebellopontine angle of a 1-year-old (Table 2).

This retrospective case series characterizes the largest series to date of BRAF V600E mutated pediatric JXG family neoplasms, all of which were first diagnosed with CNS disease and share a striking young male predominance with aggressive disease. As compared to the three previous reported BRAF mutated cases [56] and our BRAF-wild type case, there was a similar age distribution throughout, but overall more boys are represented in the BRAF V600E cohort. Radiographically, the majority of BRAF V600E CNS-JXG neoplasms had multifocal CNS disease, often with contrast enhancement and a subset were noted to have background white matter changes, suggestive of neurodegeneration, which is also a feature shared in cases of CNS-ECD and CNS-LCH [15, 45]. Two of our BRAF V600E CNS-JXG cases also presented with systemic disease, including one classic pediatric ECD with long bone involvement and one case with cutaneous JXG and associated CNS white matter disease. In both cases, the non-CNS systemic lesions also demonstrated the BRAF V600E mutation. Treatment options varied in this case series, but those with BRAF V600E may benefit from targeted inhibitor therapy, especially in aggressive or refractory disease and may halt progressive decline from histiocytosis-associated neurodegeneration, which is now recognized as a BRAF V600E driven progress [32, 43, 45]. Together with previous published cases [56], our findings support the classification of CNS-JXG neoplasms with BRAF V600E into the current “L group” histiocytic neoplasm category [21], with all CNS-JXG neoplasms benefiting from upfront molecular testing including MAPK/ERK pathway mutations and possibly also ALK fusions/mutations [12]. Thus for clinical-pathologic relevance in CNS lesions, we propose that the neuropathologist first focus on an accurate diagnosis of CNS-JXG neoplasm. Recognizing the varied histologic subtypes and shared immunophenotype with ECD is foremost. Following this, integration with molecular testing and clinical/radiographic staging will allow for a more comprehensive, integrated final diagnosis, similar to the current WHO process for other CNS neoplasms. Furthermore, recognizing malignant cytology [47] or a previous diagnosis of leukemia/lymphoma in the same patient [9], or an associated histiocytosis including LCH [38] (either concurrently with the CNS-JXG neoplasm or previously diagnosed in the same patient) is also imperative, as all three of these instances will have different and distinct outcomes. This study specifically excluded such cases, including mixed histiocytosis, which needs further investigation to understand whether BRAF V600E mixed pediatric CNS LCH-JXG lesions also share a common hematopoietic precursor, similar to adult BRAF V600E LCH-ECD histiocytosis [4, 34]. Thus, by encompassing a comprehensive diagnostic algorithm for CNS-JXG neoplasms with morphology, molecular, clinical, and radiographic correlation, the neuropathologist will enable a heightened awareness amongst the clinical team for appropriate management and treatment, including prevention of BRAF V600E driven neurodegeneration, similar to LCH [45].

The pediatric BRAF V600E CNS-JXG neoplasms in this series share histologic and variable clinical/radiographic overlap with adult ECD cases, including one classic pediatric ECD. The other BRAF mutant cases, including the systemic cutaneous case with CNS-white matter changes, are suggestion of pediatric ECD, despite no diagnostic long bone involvement or other classic radiographic ECD findings, as described in adults [15]. In fact, pediatric ECD may present differently than adults and often experience a delay in diagnosis from months to years, given the rare reporting in the literature [37‐39]. Since there are so few pediatric examples, it may be difficult to know the full clinical-radiographic spectrum of pediatric ECD, which in part may be due to underreporting in the pre-BRAF era. While in an adult, a nodular parenchymal BRAF V600E CNS-JXG diagnosed neoplasm with background CNS white matter changes and a cutaneous BRAF mutated xanthogranuloma lesion is highly suggestive of ECD [23], in children this presentation is not as well recognized as a form of pediatric ECD, especially in the pre-BRAF era [7]. In children, isolated skin JXG lesion are not known to harbor the BRAF V600E mutation (i.e., grouped as “C group” lesions) [49, 56]; however, in adults a cutaneous BRAF V600E xanthogranuloma is highly correlative with ECD, especially xanthelasmas, and should immediately prompt further clinico-radiographic investigation for ECD after biopsy diagnosis [15]. Thus, we propose that the same should be true for pediatric CNS-JXG lesions in which a morphologic diagnosis is only the first step in diagnosis. While our CNS-JXG pediatric patient, with an associated BRAF V600E skin lesion did not have classic radiographic stigmata of ECD and has thus far responded to clofarabine and dexamethasone with clinical and radiographic improvement, the background radiographic features suggestive of ECD-related neurodegeneration should be further followed in this setting. Furthermore, two other BRAF V600E positive CNS-JXG cases in our series also had features suggestive of ECD with progressive multifocal CNS disease resulting in cognitive decline, including brain atrophy. Despite the lack of long bone sclerosis or other classic ‘adult-type’ ECD findings, our cases not only share similarities with the aggressive cognitive decline that are observed in adult ECD, but also share radiographic features including associated white matter changes and brain atrophy [15, 18, 23, 29, 45].

For these reasons, adult ECD cases with CNS involvement are generally associated with poor prognosis [2]. Similarly, in one of the largest studies of previously published CNS-JXG [58], there was a higher rate (18.6%) of mortality/morbidity in both the isolated CNS-JXG neoplasms and those associated with systemic disease, as compared to the low mortality/morbidity (1–2%) of JXG in general [13, 36]. However, none of these previous JXG studies or registries included molecular testing, which would likely help further stratify patients, given our emerging data. In fact, one aggressive multifocal BRAF V600E CNS-JXG in this series, originally diagnosed in the pre-BRAF era, had poor prognosis with a rapidly progressive CNS disease with transition to hospice care, while the other BRAF V600E case, diagnosed prospectively, benefited from initiation with upfront BRAF-inhibitor therapy and had a dramatic and quick clinical response.

This type of immediate and favorable response is similar to the BRAF- and MAPK-inhibitor therapy in both adult ECD and LCH patients [17, 19, 24, 31]. However, this study was not designed to assess what constitutes best treatment protocols. It rather only highlights the lack of standard treatment protocols among the various cases. Treatment for CNS-JXG lesions should first take into account the final integrated diagnosis based on accurate morphologic diagnosis with molecular correlation and clinical/radiographic staging. However, in order to draw meaningful conclusions and develop consensus guidelines, long-term systematic study of these rare patients with follow-up is needed. To this end the Histiocyte Society’s International Rare Histiocytic Disorders Registry (NCT02285582) and subsequent prospective studies are poised to help accomplish this endeavor.

In the post-BRAF era, we now turn our attention to the molecular classification of histiocytic neoplasms as an area of ongoing, active investigation, which now includes extracutaneous JXG with BRAF V600E and MAPK pathway mutations, in addition to LCH and ECD with BRAF V600E mutations, and even rare reports of RDD with BRAF V600E [25, 44]. Thus, the question of whether the L-group histiocytic group should include only LCH/ECD or whether a more inclusive category of “MAPK-pathway activated histiocytoses” should now exist for all groups will need further discussion. Nonetheless, histology remains a seminal discriminator, as many other CNS tumors carry the BRAF V600E mutation, including both primary CNS (i.e., pleomorphic xanthoastrocytoma, ganglioglioma, pilocytic astrocytoma, papillary craniopharyngioma) and metastatic CNS tumors (i.e., melanoma, carcinomas including colorectal cancer). Thus, it is of foremost importance that the pathologist accurately diagnose these histiocytic neoplasms, with heightened awareness of their varied histopathologic patterns within the rubric of JXG family neoplasms, which may include ECD [8, 59, 60]. The radiologist must also be aware of their varied radiographic presentations as focal, multifocal, and possible association with white matter changes and brain atrophy, which can further progress years after surgical excision of the main enhancing parenchymal lesion. We advocate applying a consistent JXG-immunostain panel, including molecular based immunostains that will aid in the pathologic diagnosis of these neoplasms, given their variable morphologic features. It is also important to exclude other histiocytosis, including LCH both by morphology and CD1a/Langerin immunostains and RDD by morphology of large RDD histiocytes (with and without emperipolesis) with diffuse, dark S100/fascin immunostains [50]. At least one case in our series carried an erroneous diagnosis of RDD based on a subset of scattered S100 positive cells. Typically S100 immunostain has limited value in the CNS lesions with high background staining; however, a subset of CNS-JXG cases in this series had variable light nuclear and cytoplasmic S100 staining in the lesional histiocytes, with and without emperipolesis. This light staining pattern with S100 in a subset of JXG cells should be distinguished from CNS-RDD, which has strong/diffuse S100 and fascin staining of lesional histiocytes and lacks Factor XIIIa staining. Scattered RDD-like cells with emperipolesis and variable light S100 staining has been previously noted in cutaneous JXG-family lesions [33, 54]. Furthermore BRAF V600E mutations have also been identified in rare cases of RDD [25, 44], including a variant BRAF mutation with CNS disease [52], which further emphasizes that morphology combined with molecular are useful for accurate diagnosis.

A significant limitation of our study is the retrospective nature of this case series with limited follow-up and inability to test the BRAF-wild type cohort for additional MAPK pathway mutations. An immunohistochemical stain for phosphorylated ERK (p-ERK) is commercially available which can provide additional evidence for upregulation of the MAPK pathway as evidenced by diffuse expression in the majority of histiocytes [11, 35]. Unfortunately, many cases had no additional material to perform pERK staining. As advocated in other histiocytoses, especially those that fail standard therapy [1], the finding of MEK-ERK pathway mutations and/or upregulation by pERK may allow for more directed, targeted therapy with improved outcomes. While targeted therapy is not necessarily curative in most cases [20], it does provide a rapid and sustaining clinical response across the “L” group histiocytosis [16, 19, 32] in which there is immediate clinical response. In addition, it has value in CNS based disease not amenable to complete resection and/or in those cases that do not respond to traditional therapy protocols, including histiocytosis- associated neurodegeneration.

BRAF V600E CNS-JXG neoplasms appear enriched in male children, associated with multifocal parenchymal CNS lesions, background CNS white matter changes, and associated BRAF V600E positive systemic disease manifestations in a subset, which may in turn help expand the spectrum of pediatric ECD in the post-BRAF era. A coherent multidisciplinary approach is needed for best diagnosis, including an accurate and timely pathologic diagnosis, prospective molecular investigation, and subsequent radiographic whole-body staging to evaluate disease extent, similar to adult CNS-ECD. We propose a refinement to diagnosis of CNS-JXG based on pathology, molecular, radiology, and clinical correlation with a comprehensive diagnostic algorithm that has relevance to both clinical management and treatment protocols and is also in line with the current 2016 WHO model of reporting CNS tumors [42]. An initial morphologic diagnosis would first report histology, along with any associated results from a well-validated molecular based immunostain (i.e. BRAF VE1, pERK), if available. Only after DNA-based molecular testing with sensitive testing techniques and clinical/radiographic staging are complete should an integrated final diagnosis be rendered, with description of specific sites of involvement and molecular integration. For example, in case 3 the initial morphologic diagnosis would read as: CNS-JXG, BRAF VE1 immunostain positive. Then the final integrated diagnosis may read as: Pediatric ECD (adult-type) with involvement of brain and long bones, BRAF V600E positive. Such an integrated final diagnosis in CNS-JXG neoplasms will allow for refinement of management with tailored treatment protocols and possible expansion of the spectrum of pediatric ECD, based on pathology, molecular and clinical/radiographic correlation in the post-BRAF era.