Histone H3K36I mutation in a metastatic histiocytic tumor of the skull and response to sarcoma chemotherapy.

Recurrent somatic missense mutations in histone H3 genes have been identified in subsets of pediatric cancers. H3K36 histone mutations have recently been recognized as oncogenic drivers in rare subsets of malignant soft tissue sarcomas but have not been reported in histiocytic neoplasms. Currently, the histological and molecular spectrum, as well as the clinical behavior of H3K36-mutant soft tissue malignancies, is largely unknown. We describe a pediatric patient with a HIST1H3B K36I-mutant histiocytic tumor arising in the skull. After the failure of upfront therapy for histiocytosis and development of widely disseminated metastatic disease, the patient had an exceptional response to empiric chemotherapy and remains in complete disease remission for more than 5 years. Our report expands the histological spectrum of H3K36M/I-mutant soft tissue malignancies to histiocytic neoplasms and indicates that multiagent sarcoma-like chemotherapy can be highly effective even in the setting of widely disseminated metastatic disease.

INTRODUCTION

Somatic missense mutations in histone H3 genes have been identified as recurrent events in a subset of pediatric high-grade gliomas, and less commonly found in soft tissue and bone tumors. Lysine 36-to-methionine (K36M) mutations in H3F3B have been found in the majority of chondroblastomas (Behjati et al. 2013), and more recently, H3 mutations at lysine 36 have also been identified in two children with undifferentiated sarcomas: a HIST1H3C K36M and a HIST1H3B K36I mutation (Lu et al. 2016). The histone 3.3 K36M mutation impairs mesenchymal differentiation and has been shown to generate undifferentiated sarcomas in vivo through epigenetic reprogramming (Fang et al. 2016; Lu et al. 2016). H3K36M/I mutations lead to the reduction of H3K36 methylation and a simultaneous genome-wide gain in H3K27 methylation, resulting in a redistribution of Polycomb repressive complex 1 (PRC1) and derepression of PRC1 target genes associated with the blocking of mesenchymal differentiation.

RESULTS

Clinical Presentation, Family History, and Pathology

A 13-yr-old African–American female, with an unremarkable past medical history except for sickle cell trait, presented with a 2-wk history of recurrent progressive headaches, nausea, and vomiting. Physical examination revealed a palpable mass involving her right parietal scalp. Family history was negative for cancer and heritable disorders.

Magnetic resonance imaging (MRI) of the head with and without contrast revealed a homogeneously enhancing mass centered in the right parietal bone, with extracranial and intracranial components and associated mass effect on the parietal lobe (Fig. 1). A diagnostic skeletal survey was negative for any additional bony lesions.

Figure 1.

Initial imaging studies. (Left) Sagittal and (right) coronal contrast-enhanced, T1-weighted MR images of the head obtained at initial diagnosis, showing a homogeneously enhancing mass centered in the right parietal bone, with extracalvarial and intracranial components and associated mass effect on the parietal lobe.

An open biopsy of the skull lesion revealed an atypical histiocytic proliferation with associated T-cell infiltrate that appeared consistent with histiocytosis, although not Langerhans cell histiocytosis (Fig. 2). The lesion was composed predominantly of histiocytes that were strongly positive for CD163 and weakly positive for CD68, but negative for S100 and CD1a. The CD45-positive lymphocytic population included a CD3-positive T-cell infiltrate. A CD20- and CD79a-positive B-cell population represented only a minority of the lesion. Hematological markers ALK-1 (anaplastic lymphoma kinase), CD15, CD30, BCL-2, BCL-6, OCT-2, and PAX-5 were negative by immunohistochemistry. The tumor was negative for EBV-encoded RNA (EBER) by in situ hybridization. Soft tissue neoplasm markers α smooth muscle actin (SMA), desmin, and caldesmon were also negative. Flow cytometric analysis with a comprehensive panel of leukemia/lymphoma markers was nondiagnostic. The Ki-67 proliferation index ranged widely between various areas of the tumor, from <5% to 30% focally, with the marked inflammatory cell infiltrate component rendering it difficult to estimate the tumor cell proliferation rate.

Figure 2.

Tumor histology and immunohistochemistry. (A) Hematoxylin & eosin stain showing lymphohistiocytic neoplasm with (B) rich CD3-positive T-cell infiltrate and (C) CD163 positivity. (D) Cells show overall preserved H3K36me3 in both tumor and inflammatory cells. (E,F) In contrast, loss of H3K36me2 staining is observed in large histiocytic cells, whereas infiltrating small T cells show preserved expression. Scale bars, 100 µm.

Given the diagnostic uncertainty, a second, larger excisional biopsy was subsequently performed, and an atypical lymphohistiocytic lesion was again identified. Immunohistochemically, the atypical cells continued to be positive for AE1/AE3 (focal), vimentin, and Ki-67 (with low to moderate proliferative rate similar to the original biopsy) and negative for CD31, CD4, CD163, CD68, CD34, CD33, Factor XIIIa, CD15, CD30, CD21, CD20, CD3, Lysozyme, CD1a, S100, ALK, MPO, FLI-1, SMA, and EBV (EBER in situ hybridization).

In light of these findings, a diagnosis of Erdheim–Chester disease (ECD) was favored. The pathology was subsequently reviewed by several outside expert hematopathology and soft tissue pathology consultants, with diagnostic impressions including a nonneoplastic reactive process and malignant neoplasms including non-Langerhans cell histiocytosis, histiocytic sarcoma, and inflammatory fibrosarcoma.

Genomic and Immunohistochemistry Analysis

Whole-exome sequencing (WES) of the tumor and matched germline identified 20 somatic nonsynonymous single-nucleotide variants (SNVs) and two insertions, a frameshift insertion involving SGSM1 and a nonframe insertion involving FAM194A (Table 1). We identified a somatic mutation in HIST1H3B with an allele frequency of 20%, leading to a K36I amino acid substitution (Fig. 3). No gene fusions were identified by targeted heme and solid RNA-seq fusion panels. Immunohistochemistry for histone H3K36me2 showed decreased staining in large CD163+ tumor cells (Fig. 2) compared to the reactive T cells’ inflammatory infiltrate where H3K36me2 expression was preserved.

Table 1.

Somatic nonsynonymous mutations identified in the tumor

Gene	Chr.	HGVS DNA reference	HGVS protein reference	Variant type	Predicted effect	dbSNP	Genotype	Tumor depth	VAF (%)
ANXA9	Chr 1	c.C743T	p.A248V	Exonic	Nonsynonymous SNV		Heterozygous	314	10
FAM189B	Chr 1	c.G446A	p.G149E	Exonic	Nonsynonymous SNV		Heterozygous	290	19
WRAP73	Chr 1	c.C319T	p.L107F	Exonic	Nonsynonymous SNV		Heterozygous	115	20
ZCCHC11	Chr 1	c.C4286T	p.P1429L	Exonic	Nonsynonymous SNV		Heterozygous	234	18
KNDC1	Chr 10	c.C1708T	p.R570W	Exonic	Nonsynonymous SNV		Heterozygous	156	21
KIRREL3	Chr 11	c.C1468T	p.R490W	Exonic	nonsynonymous SNV		Heterozygous	238	23
PTPRQ	Chr 12	c.G4597A	p.V1533I	Exonic	Nonsynonymous SNV		Heterozygous	205	18
IRG1	Chr 13	c.G1384A	p.V462I	Exonic	Nonsynonymous SNV		Heterozygous	196	20
COPS2	Chr 15	c.G196A	p.E66K	Exonic	Nonsynonymous SNV		Heterozygous	374	16
CNGB1	Chr 16	c.G2540A	p.G847E	Exonic	Nonsynonymous SNV		Heterozygous	283	18
KRT39	Chr 17	c.T245A	p.L82Q	Exonic	Nonsynonymous SNV		Heterozygous	268	18
ABCA10	Chr 17	c.T6A	p.N2K	Exonic	Nonsynonymous SNV		Heterozygous	332	18
LOXHD1	Chr 18	c.G1594A	p.D532N	Exonic	Nonsynonymous SNV		Heterozygous	242	17
FOXD4L1	Chr 2	c.G61A	p.D21N	Exonic	Nonsynonymous SNV		Heterozygous	619	17
DNMT3B	Chr 20	c.G389A	p.R130H	Exonic	Nonsynonymous SNV		Heterozygous	281	19
UBE2E2	Chr 3	c.A86C	p.Q29P	Exonic	Nonsynonymous SNV		Heterozygous	190	16
ADAMTS19	Chr 5	c.A2603C	p.K868T	Exonic	Nonsynonymous SNV		Heterozygous	245	21
AK9	Chr 6	c.C2594G	p.T865S	Exonic	Nonsynonymous SNV		Heterozygous	194	16
HIST1H3B	Chr 6	c.A110T	p.K36I	Exonic	Nonsynonymous SNV		Heterozygous	138	20
PRSS16	Chr 6	c.G367A	p.G123S	Exonic	Nonsynonymous SNV		Heterozygous	157	27
SGSM1	Chr 22	c.1130_1131insC	p.D377fs	Exonic	Frameshift insertion		Heterozygous	128	10
FAM194A	Chr 3	c.163_164insAG GAGGAGG	p.V55delinsEEEV	Exonic	Nonframeshift insertion		Heterozygous	145	23

(SNV) Single-nucleotide variant.

Figure 3.

Whole-exome sequencing. Comparison of the patient's tumor and germline DNA shows somatic HIST1H3B K36I single-nucleotide mutation with an estimated variant allele frequency of 13%.

Treatment Outcome

After the second biopsy, treatment for presumed ECD with oral prednisone and intravenous cladribine was initiated (Myra et al. 2004). No response was observed and the patient developed locally progressive disease within the first two cycles of therapy (8 wk). Treatment was then changed to anakinra (interleukin 1 receptor antagonist) injections 100 mg subcutaneously once daily (Aouba et al. 2010). Anakinra was discontinued after 4 wk as a result of further disease progression, including the development of tumor-associated cerebral edema and papilledema. She subsequently underwent a radical, but subtotal resection of the skull mass. Intraoperatively, extensive bony and dural tumor involvement and invasion of the pial surface were evident.

A repeat PET/CT performed after recovery from surgery revealed widely disseminated disease with new avidly metabolic lesions involving the sternum, iliac bone, sacrum, and femur in addition to residual disease at the surgical site. Subsequently, clofarabine was initiated as salvage therapy (Simko et al. 2014) at 30 mg/m2/d for 5 d every 4 wk. After two treatment cycles, the disease progressed further, with increased hypermetabolic activity at the craniotomy site and known metastatic sites, as well as the appearance of new hypermetabolic lung nodules.

Given the lack of response to treatment regimens for histiocytosis/ECD and clinical behavior similar to an aggressive sarcoma, empiric chemotherapy with doxorubicin 75 mg/m²/cycle and ifosfamide 9 g/m²/cycle was initiated (Patel et al. 1998). This treatment led to an exceptional response, with near-disappearance of all PET avid metastatic lesions after two cycles of therapy (Fig. 4). After two additional treatment cycles with ifosfamide, involved field external beam radiation therapy was administered to the primary site, with 5580 cGy given in 31 fractions, and two additional cycles of chemotherapy with ifosfamide and doxorubicin were administered subsequently. The patient currently remains in complete disease remission for more than 5 years.

Figure 4.

Disease progression and tumor response. (A) PET/CT revealing widely metastatic tumor dissemination after 6 mo of initial therapy, with hypermetabolic lesions including the primary site as well as sternum, iliac bone, sacrum, and femur (arrows). (B) Dramatic treatment response after two cycles of chemotherapy with doxorubicin and ifosfamide, with PET/CT showing near-resolution of hypermetabolic lesions.

DISCUSSION

We report a diagnostically challenging tumor with histiocytic morphology harboring a HIST1H3B K36I mutation that did not respond to multiple histiocytosis regimens but had an excellent response to a chemotherapy and radiation therapy regimen used for soft tissue sarcomas.

The differential diagnosis in this patient was wide-ranging and comprised both benign and malignant disease processes including non-Langerhans cell histiocytosis, histiocytic sarcoma, and inflammatory fibrosarcoma. ECD is a rare non-Langerhans cell histiocytosis with multiorgan involvement and broad clinical features (Munoz et al. 2014). Diffuse histiocytic infiltration consisting of CD68-positive and CD1a-negative lipid-laden histocytes are the traditional histological hallmarks of ECD, and a high prevalence of somatic BRAF mutations (Haroche et al. 2012) has also recently been identified in this disease. ECD was initially favored in this patient based on the presence of foamy histiocytes, rare Touton-like giant cells, factor XIII positivity, lymphocytic aggregates, fibrosis, and ultrastructural features. ECD is extremely rare in children, with few cases reported to date; and the histocytic lesions are most commonly located in the long bones of the lower extremities. ECD with cranial involvement is exceedingly rare (Jain et al. 2013), and our patient did not respond to therapies previously shown to be effective in ECD.

Histiocytic sarcoma (HS) was included in the initial differential diagnosis. This rare malignant tumor of histiocytic origin is thought to arise from the monocyte/macrophage lineage and is rarely seen in children (Takahashi and Nakamura 2013). The malignant cells in histiocytic sarcoma are typically positive for CD163, CD68, and lysozyme and negative for CD1a; and our patient's tumor also lacked characteristic histological features of histiocytic sarcoma, such as the diffuse noncohesive proliferation of large pleomorphic cells with multinucleated forms.

Whole-exome sequencing of our patient's tumor revealed a single oncogenic driver alteration (i.e., a HIST1H3B K36I mutation) leading to a global reduction in H3K36me2 expression. This finding is in line with prior studies indicating that H3K36M/I mutations lead to a global reduction in H3K36me2 (Fang et al. 2016; Lu et al. 2016). To our knowledge only a single pediatric cancer patient harboring a H3K36I has previously been described, an undifferentiated thoracic sarcoma occurring in a young child (Lu et al. 2016). No data on treatment modalities and outcome has been reported in pediatric patients with H3K36M/I-mutant sarcomas. Of note, the tumor described in this report was devoid of genomic alterations in pathways frequently altered in sarcomas and associated with aggressive behavior, such as p16, p53, and RB. At present, no data exist to link the presence of a H3K36M/I mutation to responsiveness to chemotherapy. Our report expands the histological spectrum of H3K36M/I-mutant soft tissue malignancies to histiocytic neoplasms, indicating that multiagent chemotherapy can be highly effective even in the setting of widely disseminated metastatic disease.

METHODS

We performed WES of the patient's tumor and matched germline DNA, as previously described (Snuderl et al. 2018). In brief, DNA was extracted from 10 unstained 10-µm FFPE sections using automated DNA extraction Maxwell RSC DNA FFPE (Promega). Two hundred and fifty nanograms of DNA from each sample were sheared on a Covaris instrument for 360 sec (duty cycle—10%; intensity—5; cycles/burst—200). Barcoded libraries were prepared using the Kapa Low-Throughput Library Preparation Kit Standard (Kapa Biosystems), amplified using the KAPA HiFi Library Amplification kit (Kapa Biosystems) (eight cycles), and quantified using Qubit Fluorimetric Quantitation (Invitrogen) and Agilent Bioanalyzer. An equimolar pool of the four barcoded libraries (300 ng each) were used as input to capture the exome using one reaction tube of the NimbleGen SeqCap EZ Human Exome Library v3.0 (Roche, cat # 06465684001), according to the manufacturer's protocol. The pooled capture library was quantified by Qubit (Invitrogen) and Bioanalyzer (Agilent) and sequenced on an Illumina HiSeq 2500 using a paired-end, 100 nt in length run mode. The coverage was normalized by HaplotypeCaller (GATK) and the variants were called independently for each sample using MuTect2 with coverage >100× and >10% variant allele frequency (VAF) in tumors and <1% in the matched normal DNA. Resulting filtered variants were annotated using ANNOVAR RefSeq hg19. Synonymous and noncoding variants were excluded.

To assess for the presence of additional oncogenic drivers including gene fusions that may have been missed by WES, we analyzed the tumors using the MSK-Solid Fusion Assay, an RNA-based targeted sequencing panel that utilizes the Archer Anchored Multiplex PCR (AMPTM) technology and next-generation sequencing to detect gene fusions in hematologic and solid tumor samples (Zheng et al. 2014). The Archer custom heme and solid tumor panels were designed to target 200 and 62 specific genes, respectively, known to be recurrently involved in gene rearrangements. The complete gene list is provided in Table 2.

Table 2.

The Archer custom heme and solid tumor panel were designed to target 200 and 62 specific genes, respectively, known to be recurrently involved in gene rearrangements

Archer Pan Heme
ABL1*	BTK	CHD1	EBF1	HOXA10	KLF2	MYBL1	PAX5	PYRY8	SOX11
ABL2	CALR	CHIC2	EIF4A1	HOXA9	KMT2A*	MYC	PBX1	RAB29	SRSF2
AICDA	CARD11	CIITA	ENTPD1	ID4	KRAS	MYD88	PDCD1	RAG1	STAT3
AKT3	CBFB*	CREB3L2	EPOR	IDH1	LIMD1	MYH11	PDCD1LG2	RAG2	STAT5B
ALK	CBL	CREBBP	ERG	IDH2	LMO1	NEK6	PDGFRA	RANBP1	STAT6
ASB13	CCDC50	CRLF2	ETV6*	IKZF1	LM02	NF1	PDGFRB	RARA*	STIL
ASXL1	CCND1	CSF1R	EXOC2	IKZF2	LRMP	NFKB1	PHF6	RBM15	STRBP
BATF3	CCND2	CSF3R	EZH2	IKZF3	LYL1	NFKB2	PICALM	RHOA	TAL1
BAX	CCND3	CTLA4	FAM216A	IL16	LZTS1	NME1	PIMI	ROSI	TCF3
BCL11B	CD274	CYB5R2	FBXW7	IL7R	MAL	NOTCH1	PIM2	RUNX1*	TFG
BCL2	CD44	DCK	FGFR1	IRF4	MALTI	NOTCH2	PLCG1	RUNX1T1*	TLX1
BCL2A1	CD79B	DEK	FGFR2	IRF8	MAML3	NPM1	PLCG2	S1PR2	TLX3
BCL3	CDC25A	DENND3	FGFR3	ITPKB	MECOM	NRAS	PML*	SEMA6A	TNFRSF13B
BCL6*	CDK6	DLEU1	FLT3	JAK1	MKL1	NT5C2	PPAT	SERPINA9	TNFSF4
BCR*	CDKN2A	DNM2	FOXP1	JAK2	MLF1	NTRK3	PRDM16	SETBP1	TP63
BIRC3	CDKN2B	DNMT3A	FUT8	JAK3	MLLT10	NUP214	PRKAR2B	SETD2	TYK2
BLNK	CEBPA	DNMT3B	GATA1	KAT6A	MLLT4	NUP98	PTK2B	SF3B1	U2AF1
BMF	CEBPD	DNTT	GATA2	KDM6A	MME	P2RY8	PTPN1	SH2B3	WT1
BMP7	CEBPE	DUSP22	GLIS2	KIAA0101	MPL	PAG1	PTPN11	SH3BP5	XPO1
BRAF	CEBPG	E2F2	GNAS	KIT	MUC1	PAICS	PYCR1	SLC29A1	ZCCHC7
Archer Solid
ALK*	EGFR	FGFR*	JAZF1	NCOA1	NTRK3*	RET*	TFE3*
BCOR	EPC1	FOSB	KIT	NCOA2	PDGFB	ROS1*	TFEB
BRAF*	ERBB2	FOXO1*	MAML2	NOTCH1	PDGFRA	RSP02	TFG
CAMTA1	ERG*	FUS	MEAF6	NOTCH2	PHF1	RSP03	TMPRSS2
CCNB3	ETV6	GLI1	MET	NR4A3	PIK3CA	SS18*	USP6
CIC	EWSR1*	GRB7	MGEA5	NRG1	PLAGI	STAT6*	YWHAE
COL6A3	FGFR1	HMGA2	MKL2	NTRK1*	PPARG	TAF15*
DNAJB1	FGFR2*	JAK3	MYB	NTRK2	PRKACA	TCF12

*Clinically validated genes.

Briefly, RNA was extracted from tumor FFPE material followed by cDNA synthesis, cDNA end repair, dA tailing, and ligation with Illumina molecular barcode adapters. SPRI-cleaned ligated fragments are subject to two rounds of PCR amplifications using two sets of gene-specific primers (GSP1 used in PCR1 and a nested GSP2 pool that is 3′ downstream from GSP1 and used in PCR2) and a primer complementary to the Illumina adapter. At the end of two PCR steps, the final targeted amplicons were sequenced on an Illumina MiSeq instrument (2 × 150 bp). The Archer analysis software V5.0 was used for data analysis, which revealed no evidence of oncogenic gene fusions in the tumor sample.

ADDITIONAL INFORMATION

Data Deposition and Access

The HIST1H3B K36I variant described in the manuscript will be deposited in ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/) under accession number SCV000993417. Raw sequencing data was not deposited because consent for public deposition of the genomic data was not granted.

Ethics Statement

The study was conducted at NYU School of Medicine and approved by the Institutional Review Board (IRB) in accordance with all local and federal regulations. The patient provided written consent for publication; however, the consent for public deposition of the genomic data was not granted.

Acknowledgments

The study was supported in part by The Making Headway Foundation and The Friedberg Charitable Foundation and through the National Institutes of Health (NIH)/National Cancer Institute (NCI) Cancer Center Support Grant P30 CA008748 to Memorial Sloan Kettering Cancer Center.

Author Contributions

M.S. and M.A.K. conceived the study and wrote the manuscript. All authors acquired and analyzed the data and edited and approved the final version of the manuscript.

Competing Interest Statement

The authors have declared no competing interest.