Gastrointestinal stromal tumors with BRAF gene fusions. A report of two cases showing low or absent KIT expression resulting in diagnostic pitfalls

Abstract

Although most gastrointestinal stromal tumors (GISTs) exhibit activating mutations in either KIT or PDGFRA, rare cases have shown to be driven by gene fusions involving kinases, mainly involving NTRK3, and rarely BRAF or FGFR1. BRAF gene rearrangements have been described in only two patients to date, as separate case reports. In addition, BRAF V600E mutation is an uncommon but established oncogenic pathway in GIST. In this report, we describe two new GIST cases harboring novel BRAF fusion genes, arising in two young-adult women (37 and 40 years of age) in the small bowel and distal esophagus, both with a spindle cell phenotype. The small bowel GIST measured 2.8 cm and showed a high cellularity and a mitotic rate of 20/50 HPFs, while the esophageal lesion measured 7 cm and 1/50 HPFs. Immunohistochemically, both tumors showed diffuse reactivity for DOG1, while KIT/CD117 was weakly positive in the small bowel GIST and completely negative in the esophageal tumor. Based on these findings, the latter case was misinterpreted as a low-grade myxoid leiomyosarcoma, as it showed a myxoid stroma, reactivity for SMA and focal positivity for desmin. Archer FusionPlex revealed a fusion between BRAF with either AGAP3 or MKRN1 gene partners. Moreover, MSK-IMPACT DNA targeted sequencing confirmed both fusions but did not identify additional mutations. In one case with available material, the BRAF gene rearrangement was also validated by FISH. The recognition of BRAF fusion-positive GISTs is critical as it may be associated with a low level of KIT expression and may result in diagnostic challenges with significant impact on therapeutic management. The clinical benefit with KIT inhibitors, such as imatinib, remains to be determined.

1 |. INTRODUCTION

Gastrointestinal stromal tumor (GIST) is the most common mesenchymal tumor arising in the gastrointestinal tract. Over the last two decades, the molecular abnormalities underpinning these tumors have been discovered, with up to 85–92% of cases in adults harboring mutually exclusive gain-of function KIT or PDGFRA mutations.^1–3 Instead, gastric tumors in the pediatric and young adult patients are often wild-type for these mutations⁴ and harbor alterations resulting in a deficiency in the succinate dehydrogenase (SDH) complex.^2,5,6 Moreover, these SDH-deficient GISTs form the underlying pathogenetic basis of syndromic cases such as Carney triad and Carney– Stratakis syndrome.^7,8 Patients with Type I Neurofibromatosis (NF-1) also develop GIST at an increased frequency, often multifocal, typically in the small bowel and associated with interstitial cell of Cajal hyperplasia, driven by NF-1 rather than KIT or PDGFRA mutations.^9,10 Additionally, BRAF V600E mutations have been found in KIT-expressing GISTs lacking KIT/PDGFRA/SDH abnormalities, and in rare cases of imatinib-resistant GIST.^11,12 In the last few years, as a result of wide application of targeted RNA sequencing in clinical practice, a small subset of GIST driven by gene fusions resulting in oncogenic kinase activation has been identified, including a handful of cases with FGFR1 and NTRK3 fusions.^13,14 In this study, we report on two BRAF-fusion positive spindle cell GISTs which posed diagnostic challenges due to low or absent KIT expression. Thus, further investigation unmasking molecular alterations in these tumors can facilitate accurate classification and detect tumors unlikely to respond to current targeted therapy (Table 1).

TABLE 1.

Summary of clinical, pathologic, and molecular findings of four BRAF-fusions positive GIST

	Charo et al23	Vanden Bempt et al3	Case 1	Case 2
Age	34	64	40	37
Gender	F	M	F	F
Site	Small intestine	Stomach	Small intestine	Esophagus
Size (cm)	14	2.5	2.8	7.3
Morphology	Spindle	Spindle	Spindle	Spindle
Mitotic rate	3/50 HPF	≤5/50HPF	20/50 HPF	<1/50 HPF
IHC	CD117 positive DOG1 positive	N/A	CD117 weakly pos DOG1 positive 18 mo NED	CD117 negative DOG1 positive SMA positive Desmin rare pos ER negative 36 mo/NED
Fusion	PRKAR1B-BRAF PRKAR1B exon 9-BRAF exon 9	TRIM4-BRAF TRIM4 exon 6-BRAF exon 10	AGAP3-BRAF AGAP3 exon 11-BRAF exon 10	MKRN1-BRAF MKRN1 exon 4-BRAF exon 11
Survival/therapy	N/A	N/A	18 mo NED Surgery	36 mo NED Surgery

Abbreviations: mo, months; NA, not available; NED, no evidence of disease; pos, positive.

2 |. MATERIALS AND METHODS

Tissue was analyzed prospectively in the course of management as clinical cases of patients referred to our institution for continued care. The tumors were subjected to morphologic and immunohistochemical analysis, targeted RNA sequencing (Archer FusionPlex), MSK-IMPACT, and fluorescence in situ hybridization (FISH). Significant clinical follow-up is not yet available due to the recent nature of the cases. This study was approved by the institutional review board.

2.1 |. Immunohistochemistry

The immunohistochemical analysis was performed on formalin-fixed, paraffin-embedded (FFPE) tissue sections of 4 μm thickness. Antibody epitope retrieval was performed using standard protocols for the following markers: KIT (CD117), DOG1, S100, SOX10, cytokeratin AE1: AE3, desmin, and smooth muscle actin (SMA). BRAF immunohistochemistry for BRAF V600E was performed in one case with available tissue.

2.2 |. Fluorescence in situ hybridization

Fluorescence in situ hybridization (FISH) on interphase nuclei from paraffin-embedded 4-μm sections was performed applying custom probes using bacterial artificial chromosomes (BAC) covering and flanking genes of interest. A BAC clone for BRAF was chosen according to UCSC genome browser (http://genome.ucsc.edu), as previously described.^15,16 The BAC clone was obtained from BACPAC sources of Children's Hospital of Oakland Research Institute (CHORI; Oakland, CA; https://bacpacresources.org/). DNA from individual BACs was isolated according to the manufacturer's instructions, labeled with different fluorochromes in a nick translation reaction, denatured, and hybridized to pretreated slides. Slides were then incubated, washed, and mounted with DAPI in an antifade solution, as previously described.¹⁶ The genomic location of each BAC set was verified by hybridizing them to normal metaphase chromosomes. Two hundred successive nuclei were examined using a Zeiss fluorescence microscope (Zeiss Axioplan, Oberkochen, Germany), controlled by the Isis 5 software (Metasystems, Newton, MA). A positive score was interpreted when at least 20% of the nuclei showed a split-apart signal in the break-apart assay. Nuclei with an incomplete set of signals were omitted from the score.

2.3 |. Targeted RNA sequencing

RNA is extracted from formalin-fixed paraffin-embedded tumor material followed by cDNA synthesis. cDNA libraries were made using the ArcherTM FusionPlexTM standard protocol and supplied reagents, including Archer^® Universal RNA Reagent Kit for Illumina^® (Catalog #AK0040–8), as previously described.¹⁷ Fusion unidirectional gene specific primers were designed to target specific exons in 62 genes known to be involved in chromosomal rearrangements based on current literature. At the end of the two-PCR steps the final targeted amplicons were sequenced (2 × 150 bp) on an Illumina MiSeq sequencer. Data analysis was performed using the ArcherTM analysis software settings.¹⁷

2.4 |. Next generation sequencing (MSK-IMPACT)

The IMPACT next generation sequencing platform has been described in detail previously.¹⁸ It is an FDA-approved hybridization capture-based genomic sequencing assay performed in a Clinical Laboratory Improvement Amendments-certified laboratory that examines all exons and selected introns of 468 cancer-associated genes. Genomic alterations detected on IMPACT are annotated according to the OncoKB database,¹⁹ a precision oncology knowledge base denoting the oncogenic effects and predictive significance of molecular alterations. Genomic data and OncoKB annotations were visualized in cBioPortal for Cancer Genomics.^20,21

3 |. RESULTS

3.1 |. Case reports

3.1.1 |. Case 1

The patient was a 40 year-old woman who was incidentally found to have a mass involving the small intestine on imaging on a workup for a pyelonephritis. The patient underwent a laparoscopic exploration and an en-bloc resection of the proximal jejunal mass. On gross examination, the tumor measured 2.8 × 2.5 × 2.0 cm and showed a solid, white cut-surface. Microscopically, the tumor was vaguely multi-nodular and composed of intersecting fascicles of spindle cells with pale eosinophilic cytoplasm and uniform fusiform nuclei. Focally, cytoplasmic vacuoles and skeinoid fibers were observed. (Figure 1). The mitotic count was brisk (20/50 HPFs), but necrosis was not identified. Immunohistochemical analysis showed that the tumor cells were positive for DOG1 (Figure 1) and weakly positive for KIT/CD117. Based on the high-risk features, patient was started on adjuvant imatinib therapy for 1 month and then discontinued once the molecular results revealed no mutations in the KIT/PDGRFA genes. The patient is free of recurrence 18 months since diagnosis.

FIGURE 1.

Pathologic features of case 1: small intestinal GIST harboring a novel AGAP3-BRAF fusion. Low power view shows a lobulated tumor within the muscularis propria and subserosal layer of the small bowel (A). Intermediate power showing highly cellular, intersecting fascicles of spindle cells (B); while at high power show eosinophilic cytoplasm, intracytoplasmic vacuoles, and monomorphic nuclei with fine chromatin. (C). Immunohistochemically the tumor showed weak staining for KIT/CD117 (D), while there was diffuse, strong staining for DOG1 (E). (F). FISH shows break-apart red (centromeric) and green (telomeric) in keeping with a BRAF gene rearrangement (the narrow, fixed gaps between the break-apart signals support an intrachromosomal inversion)

3.1.2 |. Case 2

The patient was a 37 year-old woman who presented with dysphagia and was found to have a large mass involving the distal esophagus. The patient underwent a video-assisted thoracoscopic resection of the mass. On gross examination, the tumor appeared well-circumscribed and measured 7.3 × 5.5 × 4.0 cm, with a pale-yellow, glistening cut surface. Microscopic examination revealed a tumor with well-defined borders, compose spindle cells with fibrillary eosinophilic cytoplasm and uniform fusiform nuclei, embedded in a predominantly myxoid stroma (Figure 2). Mitotic figures were rare (1 MF/50 HPFs) and necrosis was not present. Surgical margins were free of involvement. Immunohistochemically the tumor was positive for DOG1, CD34, SMA, and focal reactivity to desmin (Figure 2). Other stains, including KIT/CD117, AE1:AE3, EMA, S100, ALK, and inhibin were negative. The patient is free of disease 36 months since initial diagnosis without further therapy.

FIGURE 2.

Pathologic features of distal esophageal GIST with MKRN1-BRAF fusion. Low power shows a well-circumscribed lesion surrounded by a fibrous capsule (A), which is composed of loose fascicles of bland spindle cells with scant eosinophilic cytoplasm and ovoid unform nuclei with fine chromatin (B,C). The tumor is associated with extensive myxoid stroma and scattered mast cells (D). Immunohistochemically the tumor cells were negative for KIT/CD117 (E) (which highlights the stromal mast cells, as internal positive control), while diffusely positive for DOG1 (F) and SMA (G), and only rare cells label with desmin (H)

3.2 |. Molecular findings

In case 1, both Archer FusionPlex and IMPACT testing showed the presence of an AGAP3-BRAF fusion gene. The Archer FusionPlex confirmed the fusion transcript involving AGAP3 exon 11 and BRAF exon 10 (Figure 3). The BRAF gene rearrangement was subsequently confirmed by a FISH study.

FIGURE 3.

Diagrammatic representation of the two intrachromosomal BRAF fusions. (A). Schematic view of BRAF gene location on 7q34 (green box) and its two fusion partners AGAP3 on 7q36.1 and MKRN1 on 7q34 (orange boxes). The direction of transcription of each gene is shown by an orange or green arrow. Green circular arrows indicate that both fusions result from a complex process of break, inversion, and fusion, resulting in a functional transcript retaining the BRAF kinase domain as the 3' partner in both cases. (B). Upper portion reveals a fusion transcript composed of AGAP3 exon 11 fused to exon 10 of BRAF; while the lower portion shows MKRN1 exon 4 fused to exon 11 of BRAF. In both cases, the projected fusion oncoprotein retains the BRAF kinase domain intact. The protein domains of the participating genes are also displayed

In case 2, both Archer FusionPlex and IMPACT testing showed the presence of an MKRN1-BRAF fusion. The Archer FusionPlex confirmed the fusion transcript involving MKRN1 exon 4 fused to BRAF exon 11 (Figure 3).

Copy number variations were assessed in case 1, where IMPACT identified copy number losses on chromosome arms 1p32, 3p11–13, 3p25, 7q36, and 14q12–31. In both cases, IMPACT did not identify any additional mutations.

4 |. DISCUSSION

Molecular analysis over the past two decades has transformed the diagnosis and management of GIST. The discovery of gain of function mutations in KIT has led to the development of diagnostic immunohistochemistry, provided prognostic indicators, and driven therapeutics.²² In the years following this initial discovery, investigators have further unraveled the molecular drivers in KIT wild-type disease to include PDGFRA/SDH(complex)/NF-1/BRAF mutations.^7–11 More recently as result of targeted RNA sequencing rare GIST cases were identified harboring oncogenic kinase gene fusions rather than kinase mutations, specifically involving FGFR1 and NTRK3.^13,14 In the study by Shi et al,¹³ among 24 wild-type GISTs for KIT/PDGFRA/ RAS mutations, two tumors harbored FGFR1 fusions involving (FGFR1-HOOK3 and FGFR1-TACC1; which included most of the FGFR1 kinase fusion domain) and one ETV6-NTRK3. In that series, the patient with small bowel GIST with ETV6-NTRK3 fusion progressed on five lines of therapy, including imatinib, sunitinib, sorafenib, nilotinib, and regorafenib, before the LOXO-101 therapy was instituted based on the genetic findings. The patient showed immediate improvement in his symptoms, with tumor response to LOXO-101 seen at the end of week 8 by PET/CT and an ongoing partial response (44%) according to RECIST 1.1 criteria. One case report of ETV6-NTRK3 fusion positive rectal GIST described by Brenca et al¹⁴ showed diffuse and strong reactivity for KIT protein by immunohistochemistry.

Recently, two GIST cases, one each harboring a PRKAR1B-BRAF and TRIM4-BRAF fusion, were reported by two groups of investigators. The PRKAR1B-BRAF fusion occurred in a 14 cm tumor in the small intestine of a 34-year-old woman,²³ while the TRIM4-BRAF fusion was identified in a 2.5 cm gastric lesion in a 64 year-old man.³ Detailed pathologic features were not reported.

We report two additional cases of GIST with BRAF gene fusions and present detailed clinical and pathologic features herein. In case 1, the tumor was located in the small bowel and exhibited a tightly packed fascicular growth of spindle cells with pale, eosinophilic cytoplasm. Focal intracytoplasmic vacuoles and skeinoid fibers were present, the latter finding usually observed in small bowel GIST.^24,25 The tumor had a high risk of malignancy based on the brisk mitotic rate. In case 2, the tumor was located in the distal esophagus and showed lobulated borders. The tumor exhibited a spindled phenotype with eosinophilic cytoplasm in an abundant myxoid stroma. The mitotic activity was inconspicuous. Despite the distinct clinical presentations, the two cases showed a similar immunophenotype, with DOG1 expression, while KIT/CD117 was either weak/focal (case 1) or completely negative (case 2). In fact, case 2 was initial misdiagnosed as a myxoid leiomyosarcoma at the primary institution, due to the abundant myxoid stroma and the lack of KIT positivity with focal desmin positivity. Positivity of smooth muscle markers is not unusual in GIST and has been previously described in GIST of the esophagus.²⁶ In contrast to SDH-deficient GIST that have a predilection for stomach and NF1-syndromic GIST with predilection for small bowel, it appears that there is no anatomic location preference in the four GIST cases harboring BRAF fusions.

The low or no KIT expression in the two BRAF-fusion positive GIST cases raises important questions regarding the histogenesis of certain molecular variants of GIST (specifically kinase fusion positive GIST) and the utility of KIT immunostaining in this setting. GISTs are believed to arise from interstitial cells of Cajal, which show high levels of KIT immunoexpression.^1,27 The majority of GISTs harbor a gain of function mutation in the KIT gene,¹ which results in constitutive activation of the KIT tyrosine kinase receptor. In a study investigating 25 KIT immunonegative, morphologically typical GISTs there were 18 tumors harboring PDGFRA mutations, four showing KIT mutations, while the remaining three tumors were considered KIT/PDGFRA wild-type.²⁸ Moreover, two other molecular variants of GIST, including SDH-deficient GIST²⁹ and BRAF V600E-mutant GIST¹² also show diffuse positivity for KIT. However, the status of KIT immunoexpression in the context of kinase-fusion positive GISTs is less clear. Intriguingly, in the study by Shi et al,¹³ the status of KIT/DOG1 expression in the GIST with FGFR1 and NTRK3 fusions was not provided, while the single case report in the study by Brenca et al14 of a rectal GIST with ETV6-NTRK3, both KIT and DOG1 were diffusely and strongly positive by immunohistochemistry. In contrast, a recent report of 8 mesenchymal tumors of the GI tract with NTRK gene rearrangements (including both NTRK1 and NTRK3 fusions) showed none were KIT or DOG1 immunopositive and thus appeared to be unrelated to GIST.³⁰

Moreover, the status of KIT/DOG1 staining was only documented in the small bowel GIST with PRKAR1B-BRAF fusion²³ being both positive, while no information was provided in the second case of a gastric GIST.³ Although the weak or absent KIT immunoreactivity in our two cases with BRAF fusions triggered diagnostic challenge, the diffuse DOG1 staining combined with morphologic appearance confirmed the correct diagnosis of GIST. Similarly, DOG1 immunohistochemical stain has proven to be a reliable marker of KIT-negative GIST, particularly in the setting of PDGFRA mutant GISTs or other unusual molecular GIST subtypes.^31,32

BRAF is a serine/threonine kinase and a member of the RAF family. Alterations involving the BRAF gene are increasingly recognized in human neoplasia.³³ The point mutation resulting in the BRAF V600E mutant is present in 3.9–13% of GIST lacking KIT/PDGFRA mutations.^11,12,34 BRAF related fusions have been previously described in other mesenchymal neoplasms, including infantile fibrosarcoma-like tumors³⁵ and myxoinflammatory fibroblastic sarcoma.³⁶ In other cancers, BRAF fusions encode 5' protein partners that contribute coiled-coil or zinc-finger dimerization motifs, which likely produce constitutively activated BRAF dimers capable of driving tumorigenesis and poorly sensitive to RAF inhibitors, but sensitive to inhibition downstream, through MEK (mitogen-activated protein kinase kinase 1 and 2) inhibition.

Finally, the most critical impact of the BRAF fusion alteration in GISTs is the predicted drug resistance to specific tyrosine kinase inhibitor therapy, such as imatinib, the front line targeted therapy in metastatic or locally advanced GIST.²² Equally important are the alternative strategies which can be offered in the setting of this genotype, such as targeted therapies including RAF/pan-kinase inhibitor therapy (sorafenib)^37,38 and MEK inhibitor therapy.³⁹

In summary, we describe two cases of GIST with unusual morphologic and immunohistochemical findings and underlying BRAF related gene fusions. Awareness of this molecular variant of GIST may support the use of broader molecular analysis to improve diagnostic accuracy and broaden the scope for targeted therapy.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.