Recurrent CLTC::SYK fusions and CSF1R mutations in juvenile xanthogranuloma of soft tissue Free

Xanthogranuloma (XG) family lesions are a group of histiocytic diseases that usually present in the skin.¹ Various subtypes can be distinguished based on the clinical setting and histopathologic features.¹ In children, “juvenile XG (JXG)” or “JXG family lesions” are used as umbrella terms for these subtypes, which have overlapping histomorphology and a shared immunophenotype.^2,3 The term JXG is used in the World Health Organization (WHO) classification of hematolymphoid tumors.⁴ Most JXG lesions appear during the first years of life,⁵ and skin lesions often resolve spontaneously within months to years.¹ Rarely, extracutaneous localizations of JXG occur, which may pose a diagnostic challenge and can behave more aggressively.^2,3,6 Recent studies have identified somatic genetic alterations in many histiocytic diseases,^7-12 including JXG. However, the genetic drivers of the rare form of extracutaneous JXG remain incompletely characterized. Here, we demonstrate targetable somatic genetic alterations in extracutaneous JXG, including recurrent CLTC::SYK fusions and CSF1R mutations in children below 2 years of age with large soft tissue tumors.

We searched for children diagnosed with extracutaneous JXG between 1971 and 2022 in the Dutch Nationwide Pathology Databank (Palga).¹³ In addition, we searched for adults with isolated extracutaneous XG family lesions, a disease presentation that does not fit with the clinical phenotype of Erdheim-Chester disease (ECD).¹⁴ Notably, the term “juvenile” in JXG only indicates a XG family lesion in a child (aged <18 years at diagnosis); JXG, adult XG, and ECD share similar morphologic patterns and a common immunophenotype.² First, we retrieved all pseudonymized pathology reports with a pathologist-assigned diagnostic code for a histiocytic disorder.^5,15 Relevant cases were then identified based on free-text queries (directed at terminology for organs other than the skin), because no separate diagnostic code for extracutaneous JXG exists.¹⁵ Archival tissue slides and formalin-fixed paraffin-embedded (FFPE) tissue blocks were requested from pathology laboratories through the “Palga intermediary procedure.”¹⁶ In addition, the pathology laboratories referred us to the treating physicians, who provided pseudonymized clinical data and images. Central pathology review was performed by an experienced soft tissue pathologist (P.C.W.H.), and additional immunohistochemical stains were performed when indicated. Inclusion criteria were: (1) XG morphology and immunophenotype, as detailed in the WHO classification of hematolymphoid tumors (chapter JXG),⁴ with expression of at least 1 of the histiocyte markers CD68 and CD163; (2) the presence of at least 1 extracutaneous disease manifestation, which encompasses subcutaneous/soft tissue lesions (sometimes referred to as “deep” or “giant” JXG lesions^17-20); and (3) sufficient leftover tissue for molecular analysis - unless comprehensive molecular analysis had already been performed through routine diagnostics. Excluded were cases diagnosed as ECD based on characteristic clinical and radiologic features¹⁴ or as ALK⁺ histiocytosis, malignant histiocytosis, or mixed histiocytosis based on pathologic findings.^4,9,21 Antibody clones and immunohistochemistry protocols are provided in the supplemental Methods, available on the Blood website. This retrospective study was approved by the Palga Scientific Council and Privacy Committee (LZV-2016-183) and the institutional review board of Leiden University Medical Center (B19.074); the requirement for informed consent was waived. In addition, the study was approved by the Biobank and Data Access Committee of the Princess Máxima Center for Pediatric Oncology (PMCLAB2024.0512).

Unless cases had been comprehensively analyzed through routine molecular diagnostics, DNA was isolated from microdissected FFPE tissue fragments or from FFPE tissue punches and subjected to targeted locus capture–based next-generation sequencing (TLC-NGS) according to a published protocol.^10,22,23 TLC-NGS leverages DNA crosslinking and fragmentation and is therefore particularly suitable for analyzing (old) FFPE samples in which DNA is inherently crosslinked and fragmented.²³ Moreover, TLC-NGS enables the detection of both small genomic variants and chromosomal rearrangements. For this study, a custom capture probe panel (Roche) was designed, capturing all exons of 126 genes frequently mutated in histiocytic and other hematologic neoplasms, as well as introns of 55 of 126 genes. A list of targeted genes is provided in the supplemental Methods. NGS reads were aligned to the human GRCh37/hg19 reference genome before the detection of gene fusions and other structural variants using the computational framework proximity ligation–based identification of rearrangements (PLIER).^22,23 To detect small genetic variants, sequencing data were subjected to a published somatic variant calling pipeline,²⁴ using variant caller Mutect2.²⁵ Variants were analyzed by a certified clinical molecular biologist in pathology (T.v.W.) using Geneticist Assistant software (SoftGenetics), as described previously.^10,26 

For polymerase chain reaction (PCR) and Sanger sequencing, DNA was isolated from FFPE tissue using the automated Tissue Preparation System (Siemens Healthcare Diagnostics).²⁷ Primers spanning the fusion break points were designed with Primer3 software²⁸ and obtained from Integrated DNA Technologies. Primer sequences and detailed PCR and Sanger sequencing methodology are available in the supplemental Methods. In short, PCR was performed on a Bio-Rad CFX96 Real-Time PCR Detection System with SYBR Green Supermix (Bio-Rad Laboratories), with an annealing temperature of 60°C for 40 cycles. The PCR product was purified using the QIAquick PCR Purification Kit (Qiagen) and subjected to Sanger sequencing on an Applied Biosystems 3730XL system.

For SYK break-apart fluorescence in situ hybridization (FISH), bacterial artificial chromosome clones were obtained from the Wellcome Sanger Institute. One bacterial artificial chromosome clone proximal of SYK (RP11-555F9, chr9: 93146180-93322501, GRCh37/hg19) and 1 distal of SYK (RP11-440G5, chr9: 94132889-94301786, GRCh37/hg19) were used.²⁹ The proximal probe was labeled with digoxigenin-11-dUTP and the distal probe with biotin-11-dUTP (Roche), as described earlier.³⁰ FISH was then performed on 6-μM thick FFPE tissue sections according to previously described methodology with minor modifications.^30,31 Details are available in the supplemental Methods.

Whole transcriptome sequencing was performed in the routine workflow of the Princess Máxima Center for Pediatric Oncology.³² Gene fusions were detected with STAR-Fusion software.³³ 

Immunohistochemistry was performed on 3-μm thick FFPE tissue sections using commercially available antibodies for spleen tyrosine kinase (SYK), phosphorylated-SYK (p-SYK), p-extracellular signal-regulated kinase (p-ERK), p-Akt, p-ribosomal protein S6 (p-S6), and cyclin D1. In addition, multispectral immunofluorescence staining of CD163, SYK, and p-SYK was performed; antibody clones and protocols are available in the supplemental Methods.

Genome-wide methylation profiling and copy number analysis was performed using the Infinium MethylationEPIC v2.0 array (Illumina) in the routine workflow of Heidelberg University Hospital.³⁴ In addition to 15 tissue samples of patients with extracutaneous JXG, we profiled 14 FFPE samples of patients with isolated cutaneous XGs. As control, we profiled 10 FFPE samples of common dermatofibromas and 8 FFPE samples of atypical dermatofibromas.³⁵ Unsupervised hierarchical clustering was performed using the 20 000 most variably methylated 5'-C-phosphate-G-3' (CpG) sites across the data set according to standard deviation, Euclidean distance, and the complete linkage method. Details of data analysis are provided in the supplemental Methods.

Sixteen children with extracutaneous JXG were identified (Table 1), with a median age at diagnosis of 4.5 months (range, 0-2.8 years). Thirteen had isolated soft tissue tumors, 1 had central nervous system (CNS)–JXG, and 2 had systemic JXG affecting ≥2 organ systems. In addition, 5 adults with XGs confined to the CNS or soft tissue were included, with a median age of 47 years (range, 18-58). Patients were diagnosed between 1987 and 2022 at 10 different hospitals across The Netherlands; the median follow-up was 4.3 years (range, 0-25). All 13 children with soft tissue tumors were below 2 years old; their tumors manifested in diverse anatomic locations, including the scalp, trunk, extremities, and trachea. Both children with systemic JXG did not meet diagnostic criteria for pediatric ECD.³⁶ 

CSF1R alterations were detected in 7 children, including 6 males and 1 female (Table 1). Five had a deletion in exon 12 of CSF1R (NM_005211.3), whereas 1 patient had multiple missense mutations in exons 9 and 10, and the female patient had a large deletion of exons 21 and 22, in addition to an MRC1::PDGFRB fusion. The 5 children with exon 12 deletions were between 1 and 4 months old at the time of histopathologic diagnosis; 3 of 5 had tumors at birth. The tumors were up to 4 cm large, with occasional ulceration and a remarkable nodular aspect of some lesions (Figure 1). Two patients had multifocal tumors, including case 3 with a neck and upper leg lesion (Figure 1B) and case 7 with a subcutaneous scalp lesion and intramuscular tumor on the back. In case 5, the subcutaneous tumor extended inward between the forearm muscles (Figure 1C). Most tumors were resected surgically, although spontaneous regression of lesions was observed in 2 patients (Figure 1B-C). At last follow-up, 3 of 5 patients were in complete remission (at 2-11 years after diagnosis), whereas 2 of 5 still had active disease.

The 2 patients with alternative CSF1R mutations had distinctive clinical features. Case 13 with multiple mutations in exons 9 and 10 was a 1-year-old male with an endotracheal pedunculated tumor. The tumor was resected, without evidence of local recurrence or distant spread at 2.2 years of follow-up. Case 6 with a large deletion of CSF1R exons 21 and 22 and an additional MRC1::PDGFRB fusion was a 3-month-old female with a large intra-abdominal tumor involving the greater omentum. In addition, multiple small lesions on the intestinal walls and in the liver hilum were observed during laparoscopy. Subtotal surgical resection was performed, and the patient achieved complete remission without disease relapse during 24 years of follow-up.

Histopathologic features of CSF1R-mutated tumors were those of the JXG family (Figure 2).^1,2 Lesions were characterized by dense infiltrates of cells with round to ovoid or sometimes elongated nuclei, often conspicuous nucleoli and eosinophilic cytoplasm. Mitotic rates were low, and cytological atypia was not observed. Touton giant cells were abundant in some (cases 1, 3, and 13), occasional in others (cases 4-5) and absent in 2 of 7 (cases 6-7). A storiform-like pattern was noted in cases 4 and 6 (Figure 2). The number of xanthomatous cells varied between patients and samples; for example, the intramuscular tumor on the back in case 7 contained large collections of foamy histiocytes, whereas the scalp lesion of this patient consisted of a dense monomorphic infiltrate without lipidized histiocytes (Figure 2). Collections of foamy histiocytes were also observed in case 6 with a CSF1R deletion and an MRC1::PDGFRB fusion (supplemental Figure 1). Tumor cells stained positive for CD68, CD163, and PU.1 (supplemental Table 3).

CLTC::SYK fusions were detected in 6 children, including 3 males and 3 females (Table 1). Patients were between 2 and 12 months old at the time of histopathologic diagnosis; 1 of 6 had a noticeable tumor at birth. All children had unifocal soft tissue tumors, which were up to 7 cm large and were generally characterized by no or limited superficial skin changes (Figure 3). The tumors were frequently located on or just above the fascia of a muscle, although case 9 had an intramuscular tumor. In case 10, the parasternal mass extended between the ribs to the parietal pleura (Figure 3). Five patients were treated surgically, whereas case 2 was managed by active monitoring and experienced spontaneous regression of the tumor (Figure 3). At last follow-up, all 6 patients had no evidence of disease nor had experienced a relapse, including 3 patients with long-term follow-up (12 to 25 years).

Similar to the CSF1R-mutated neoplasms, the overall morphologic and immunophenotypic features of tumors with CLTC::SYK fusions were those of the JXG family (Figure 4). Again, lesions were characterized by dense infiltrates of cells with round to ovoid nuclei and sometimes more spindled morphology. Mitotic rates were low, and cytological atypia was not observed. In cases 9 and 11, a storiform-like pattern and histiocytes dissecting through the musculature were noted (Figure 4). Xanthomatous histiocytes were dominant, with collections of foamy cells in multiple samples (Figure 4). Eosinophilic granulocytes were frequent in some tumors (cases 2 and 9) but rare in others. The main difference with CSF1R-mutated neoplasms was that tumors with CLTC::SYK fusions generally lacked Touton giant cells, with only a rare Touton-type giant cell observed in case 12 (supplemental Figure 2). The absence of Touton giant cells was previously noted by Crowley et al in 3 cases³⁸ and by Glembocki et al in a single case³⁹; the clinicopathologic features of these 4 cases and our 6 patients with CLTC::SYK fusions are summarized in Table 2. Tumor cells stained positive for CD68, CD163, PU.1, and CD14, whereas CD4 was often more weakly positive and OCT2 weak to negative (Table 2; Figure 4). Negative stains included S100, CD1a, cytokeratin AE1/3, and smooth muscle actin.

SYK fusions detected by TLC-NGS were validated by break point–specific PCR and Sanger sequencing in 6 of 6 cases (Figure 5; supplemental Figure 3). In addition, we confirmed the presence of the translocation in the lesional cell population using SYK break-apart FISH, which was interpretable in 4 of 6 cases. Genomic break points of the CLTC::SYK fusions were located in either exon 5 or intron 5 of SYK. Whole transcriptome sequencing of the tumors of cases 9, 10, and 12 demonstrated that break points in exon 5 or intron 5 both lead to fusion of CLTC exon 31 to SYK exon 6 at the messenger RNA level. Thus, exon 5 break points seem to result in alternative splicing through exon skipping.

Tumors with CLTC::SYK fusions strongly expressed p-SYK, indicative of SYK activation (Figure 5E). Using multispectral immunofluorescence, p-SYK expression was demonstrated in CD163⁺ histiocytes rigorously (Figure 5F). The tumor cells also stained positive for cyclin D1 and p-S6, a marker of mammalian target of rapamycin (mTOR) activation.^40,41 Moreover, p-ERK expression was observed in both mononucleated and multinucleated cells in cases 11 and 12, whereas rare p-ERK⁺ cells were seen in the other 4 cases (supplemental Figure 2). p-Akt immunohistochemistry was consistently negative.

Genome-wide methylation profiling revealed that histiocytic tumors with CLTC::SYK fusions did not have a distinct methylation profile from other XG family lesions (Figure 5G). Although XG family lesions clustered separately from common and atypical dermatofibromas (control), there was no segregation of molecular and/or clinical subgroups within the XG family lesions. More specifically, both CSF1R mutations and CLTC::SYK fusions were present in each of the 2 major clusters identified within the XG family lesions. Considering the pronounced similarities between these 2 clusters, the biological significance of their distinction is questionable; instead, one could also deduce that XG family lesions have a common methylation profile. Furthermore, tumors with CLTC::SYK fusions did not have a distinct copy number signature, as they displayed flat copy number profiles like CSF1R-mutated lesions (supplemental Figure 4).

The 4 adults were between 18 and 58 years old at diagnosis and had unifocal tumors in the nasopharynx (case 14), paranasal sinus (case 15), pubic region (case 16) or scalp (case 17) that were surgically resected. The TBL1XR1::BOD1L1 fusion in case 14 was detected on both the DNA and RNA levels, using TLC-NGS, FISH, and whole transcriptome sequencing. In addition, the reciprocal BOD1L1::ABHD10 fusion was detected by TLC-NGS. The tumor had XG histology with abundant Touton giant cells and collections of foamy histiocytes (supplemental Figure 5) and a flat copy number profile (supplemental Figure 6). The tumors in cases 15 and 16 also contained Touton giant cells, whereas the scalp lesion in case 17 was mostly characterized by mononucleated foamy histiocytes, with remarkable extension into the underlying bone.

The 2 patients were a 5-month-old girl with an intramedullary spinal cord tumor (case 18) and a 50-year-old woman with multifocal tumors in the brain and spinal cord (case 19). Complete systemic assessments revealed no lesions outside the nervous system. Both patients harbored BRAF^V600E and were successfully treated with BRAF inhibitors, as described in detail elsewhere.⁴² The resected cerebellar tumor of case 19 exhibited XG histology with abundant Touton giant cells (supplemental Figure 7). In case 18, mostly bland histiocytes and rare multinucleated giant cells with a ring of nuclei, suggestive of a Touton giant cell, were seen (supplemental Figure 7). The histology was somewhat atypical, with an elevated Ki67 proliferation index (30-40%) and rare mitotic figures (up to 1-2 per 10 high power field). However, no atypical mitoses or other histologic features supporting a diagnosis of malignant histiocytosis were observed; therefore, a diagnosis of CNS-JXG was made. The tumor was found to harbor multiple copy number alterations (supplemental Figure 6), which may have contributed to the atypical histologic features.

The 2 children included 1 female infant with a single skin lesion and congenital anemia, thrombocytopenia, and hepatosplenomegaly who required multiple transfusions (case 20) and a 2-year-old girl with a thickened pituitary stalk, bilateral lung lesions, enlarged mediastinal lymph nodes, and papular skin lesions (case 21; supplemental Figure 8). Spontaneous disease regression was observed in both patients, although this took multiple years in case 21 (supplemental Figure 8). Touton giant cells were abundant in case 20 and absent in case 21 (supplemental Figure 8). In both, mitotic rates were low, cytological atypia was not observed, and flat copy number profiles were obtained (supplemental Figure 6).

We present detailed clinicopathologic and molecular data of 16 children with extracutaneous JXG and 5 adults with XGs confined to the CNS or soft tissue. Using an innovative sequencing technique, we identified recurrent CLTC::SYK fusions and CSF1R mutations in children below 2 years of age with soft tissue tumors. In 1 child, we identified both a CSF1R deletion and an MRC1::PDGFRB fusion. In addition, we detected targetable alterations of BRAF, MAP2K1, or NTRK1 in patients with CNS-XG or systemic JXG.

CSF1R encodes for colony-stimulating factor 1 receptor (CSF-1R), a receptor tyrosine kinase involved in monocyte and macrophage development.^43-46CSF1R mutations have previously been described in patients with different histiocytic neoplasms, but clinicopathologic data were often scarce.^8,11,47-49 Most reported JXG cases had CSF1R exon 12 indels,^8,11 which were also detected in 5 of our cases and have been rarely described in adults with ECD or malignant histiocytosis.^8,47,49 These mutations affect the intracellular, autoinhibitory, juxtamembrane domain of CSF-1R, leading to constitutive activation of the receptor.^8,49 Activation of CSF-1R induces signal transduction through the MAPK and PI3K-Akt-mTOR signaling pathways,^8,49 with preferential activation of the latter.⁴⁹ The activation of multiple downstream signaling pathways has therapeutic implications, as illustrated by a recent case report of a woman with CSF1R-mutated ECD who did not respond to the MEK inhibitor cobimetinib but did have a complete response to the CSF-1R inhibitor pexidartinib.⁴⁹ None of our patients required systemic therapy, and cases 3 and 5 demonstrate that lesions may regress spontaneously (Figure 1). Thus, treatment decisions should be taken with prudence, avoiding mutilating surgery in these young children.

The CSF1R missense mutations in case 13 were mostly located in exon 10, wherein a deletion has previously been detected in a child with LCH.⁸ Exon 10 mutations affect the extracellular region of CSF-1R and might enhance receptor dimerization.⁸ In contrast, the large deletion of CSF1R exons 21 and 22 in case 6 affects the intracellular c-CBL binding domain.⁵⁰ Binding of c-CBL to CSF-1R leads to receptor ubiquitination, which targets it for degradation.^50,51CSF1R exon 21 or 22 mutations have been described in different histiocytoses and likely prevent the c-CBL–mediated degradation of CSF-1R.^8,11,48 As in our case, CSF1R exon 21/22 mutations were sometimes accompanied by another kinase alteration.⁴⁸ Durham et al already demonstrated that expression of a CSF1R exon 10 or exon 22 mutation in Ba/F3 cells conferred cytokine-independent growth, supporting their pathogenicity.⁸ 

CLTC::SYK fusions were only recently described in histiocytic neoplasms.^38,39 All 4 patients in the literature were below 2 years old and had isolated soft tissue tumors, just as our 6 cases. Although Crowley et al considered the tumors to be histopathologically distinct from JXG, we and Glembocki et al advocate that the overall morphologic and immunophenotypic features are those of the JXG family, despite the general absence of Touton giant cells. Notably, Touton giant cells are not an essential diagnostic criterium for JXG.⁴ Accordingly, Touton giant cells were also absent in other JXG lesions, such as those from cases 6 and 7. In addition, we observed striking histologic similarities between histiocytic lesions with SYK fusions and those with CSF1R mutations (Figures 2 and 4). The tumors with SYK fusions also lacked a distinct methylation profile and exhibited flat copy number profiles like CSF1R-mutated lesions. Finally, the clinical course was characteristic of JXG, with spontaneous regression of lesions or complete remission after (sub)total resection. Therefore, we regard histiocytic tumors with CLTC::SYK fusions as a molecular subtype of the JXG family.

SYK is a nonreceptor tyrosine kinase that is primarily expressed in hematopoietic cells, including B cells, monocytes, and macrophages.⁵² SYK has diverse biological functions, including mediating signals from B cell receptors, Fc receptors, and adhesion receptors.⁵² SYK and CSF-1R are also connected, as SYK can be activated by CSF-1R signaling through the intermediary DAP12 protein.⁴⁴ The kinase domain of SYK is held in an inactive confirmation by the Src homology 2 (SH2) domains.⁵³ The CLTC::SYK fusion results in the deletion of these SH2 domains (Figure 5B) and thereby in a continuously active kinase. CLTC encodes for the clathrin heavy chain; the CLTC::SYK fusion could position the SYK kinase domains at the inside of a clathrin sphere, where they could be protected from proteasomal degradation while still being accessible to their substrate (supplemental Figure 9).

SYK fusions have been described in different hematologic neoplasms, including ITK::SYK in follicular T cell lymphoma^54,55 and ETV6::SYK in B cell acute lymphoblastic leukemia (B-ALL)⁵⁶ and atypical myeloid neoplasms with eosinophilia and histiocytic skin lesions.^57-60 Similar to CLTC::SYK, these fusions result in a transcript joining a gene fusion partner to SYK exon 6.^56-58 Expression of these SYK fusions in Ba/F3 cells conferred cytokine-independent growth and activation of SYK and its downstream targets, including the MAPK and PI3K-Akt-mTOR pathways.^57,58,61,62 Although we observed p-ERK expression in some of our cases (supplemental Figure 2), a consistent lack of p-Akt expression was noted in tumors with CLTC::SYK fusions. All tumors stained positive for p-S6, suggesting that Akt-independent mechanisms of mTOR activation may be at play.^63,64 Activation of mTOR is targetable, with clinical efficacy of the mTOR inhibitor sirolimus already demonstrated in ECD⁶⁵ and in some cases of JXG.⁶⁶ Furthermore, all tumors stained positive for cyclin D1, which is an emerging marker of neoplastic histiocytes.^9,21,67-71 Cyclin D1 expression is generally attributed to MAPK pathway activation,^67,69 although strong expression has also been observed in cases without MAPK pathway mutations and/or lack of p-ERK expression.^21,70,71 MAPK pathway inhibition is often effective in histiocytic neoplasms,^72-74 and the ETV6::SYK fusion appears to confer partial susceptibility to MEK inhibition.⁵⁹ Thus, tumors with CLTC::SYK fusions may also respond to this targeted therapy. Finally, several oral SYK inhibitors have been developed, such as fostamatinib and entospletinib.⁷⁵ETV6::SYK-transformed cells were sensitive to SYK inhibition in vitro,⁵⁸ and a patient with an ETV6::SYK–rearranged myeloid neoplasm experienced a partial response to fostamatinib.⁶⁰ Although none of our patients with CLTC::SYK fusions required systemic therapy, the discovery of activating SYK fusions opens novel opportunities for targeted therapy when necessary.

The MRC1::PDGFRB fusion detected in case 6 has been previously described in a child with JXG of soft tissue³; therefore, we establish it as another recurrent genetic abnormality in extracutaneous JXG. Strikingly, this patient from the literature had a chemotherapy-refractory left chest wall mass and was treated with dasatinib.³ In contrast, subtotal surgical resection was performed in our case, without disease recurrence after 24 years of follow-up. Thus, the clinical course of patients with identical kinase alterations can be variable. The significance of the TBL1XR1::BOD1L1 (and reciprocal BOD1L1::ABHD10) fusion in case 14 is uncertain. Although TBL1XR1 loss-of-function mutations have been described in lymphomas⁷⁶ and the gene is a recurrent fusion partner,^77-80 the fusion with BOD1L1 has not been previously described. BOD1L1 is a genome stability factor,⁸¹ whereas ABHD10 is a mitochondrial protein⁸²; neither has been implicated in oncogenic gene fusions. When another tumor harboring this fusion is identified, functional studies are warranted. Finally, BRAF^V600E is an established driver of CNS-XG,² and MAP2K1 and NTRK1 are recurrently altered in histiocytic neoplasms.^7,8,83-85 Activating alterations of these genes are targetable, with robust and durable responses to BRAF inhibition in both of our patients with CNS-XG.⁴² For systemic JXG, therapy is not always needed, as illustrated by our 2 cases.

Although we identified cases from a nationwide pathology databank, we cannot rule out some selection bias limiting our study. Because no separate diagnostic code for extracutaneous JXG exists, we were reliant on free-text queries of retrieved pathology reports. It is remarkable that we did not identify cases with BRAF fusions, because these rearrangements have been previously identified in (extracutaneous) XG family lesions.⁸⁶ Furthermore, staging was performed according to standard-of-care practices at local institutions; therefore, not all patients were systematically evaluated using the same imaging techniques (eg, positron emission tomography). Yet, both patients with ECD-associated BRAF^V600E mutations did not have extracutaneous lesions by magnetic resonance imaging (case 18) or positron emission tomography (case 19). Finally, we used a targeted sequencing approach (TLC-NGS); therefore, we might have missed relevant genetic alterations, for example in cases 15 to 17. More comprehensive molecular techniques are probably required to uncover (novel) genomic drivers in such cases. With TLC-NGS, however, we could analyze up to decades-old FFPE samples of this rare neoplasm.

In conclusion, we identify targetable kinase alterations in extracutaneous JXG, including recurrent CLTC::SYK fusions and CSF1R mutations. Supported by detailed clinicopathologic data, our study advances the molecular understanding of this rare disease and may provide guidance for the diagnosis and clinical management of patients.

The authors thank the immunolaboratories of the Departments of Pathology of Leiden University Medical Center and the Princess Máxima Center for Pediatric Oncology for performing additional immunohistochemical stains and M. J. Koudijs (Princess Máxima Center for Pediatric Oncology) for performing RNA sequencing. The authors also thank the Departments of Pathology of Maastricht University Medical Center, Haga Hospital, and Deventer Hospital for providing individual tissue samples; J. C. Jansen, J. P. W. Don Griot, M. Donker, and J. H. Allema for providing clinical information and/or images of individual patients. Finally, they thank the patients and/or their parents for allowing the use of medical photographs.

This work was supported by a grant from Stichting de Merel (P.G.K., T.v.W., A.G.S.v.H., and P.C.W.H.). P.G.K. received an MD/PhD grant from Leiden University Medical Center.

Contribution: P.G.K., A.G.S.v.H., and P.C.W.H. designed the study; H.J.B. and K.S. performed and/or supervised fluorescence in situ hybridization experiments; R.H.P.V. subjected the sequencing data to the published somatic variant calling pipeline²⁴; M.H.L. performed the in silico structural analysis of the CLTC::SYK fusion; B.K. helped with multispectral immunofluorescence imaging; I.H.B.-d.B. supervised immunohistochemistry; M.A.-H. provided tissue samples of common and atypical dermatofibromas for methylation profiling; S.W.L., J.V.M.G.B., A.H.G.C., R.M.V., C.J.M.v.N., M.R.v.D., M.A.S.-V., and P.C.W.H. were involved in the routine pathologic evaluation of included patients and/or provided archived tissue samples; P.C.W.H. performed the central pathology review; A.H.B. helped with selecting the Palga search strategy and assisted in the Palga intermediary procedure; J.A.M.v.L., A.C.H.d.V., W.J.E.T., and C.v.d.B. were involved in the clinical care of patients and provided pseudonymized clinical data and images; J.F.S. and E.S. assisted in targeted locus capture–based next-generation sequencing (TLC-NGS) panel design, performed TLC-NGS, and analyzed sequencing data for structural variants; A.v.D. coordinated the DNA methylation profiling; T.v.W. reviewed genetic variants called by the variant calling pipeline; P.G.K. collected all information, performed experiments, made the figures and tables, and drafted the manuscript; T.v.W., A.G.S.v.H., and P.C.W.H. revised the manuscript; all authors have reviewed and approved the manuscript before submission.

Conflict-of-interest disclosure: J.F.S. and E.S. are employees of Cergentis BV (a Solvias company). The remaining authors declare no competing financial interests.

Correspondence: Paul G. Kemps, Department of Pathology, Leiden University Medical Center, PO Box 9600, 2300 RC Leiden, The Netherlands; email: p.g.kemps@lumc.nl.