Whole-Exome Sequencing of Salivary Gland Mucoepidermoid Carcinoma Free

Mucoepidermoid carcinoma (MEC) is the most common malignant neoplasm arising from the salivary gland (1). MEC is characterized by its cellular heterogeneity and consists of mucin-producing, epidermoid, and intermediate cells. These tumors are also known for their variable biological and clinical behavior (2). Several histologic grading systems have been proposed, which classify MEC tumors into low, intermediate, and high grades and have been shown to correlate with clinical outcome. However, issues with consistency and reproducibility in histologic grading still exist (3, 4). Furthermore, systemic treatment options remain limited, necessitating a deeper understanding of the molecular underpinnings of this cancer.

One important genetic aberration in MEC is the translocation of chromosomes 11q and 19p, which has been proposed as an early event in the pathogenesis of the disease (5, 6). This translocation, reported in over 50% of MEC tumors (7), results in fusion of the MECT1 and MAML2 genes, forming a fusion protein that causes disruption of cell-cycle regulation and differentiation (8). Low-grade tumors have a higher incidence of the fusion compared with high-grade tumors (9), and patients with fusion-positive cancer tend to have improved survival, with significantly lower risk of local recurrence, metastases, or cancer-related mortality (10).

Other genetic alterations have been identified in MEC. For example, copy-number variations (CNVs) have been found to occur more frequently in fusion-negative cancers, with potential loss of tumor suppressor genes, such as DCC, SMAD4, GALR1, and CDKN2A/B, and gain of oncogenes, such as MAFA, LYN, MOS, and PLAG1 (9).

Comprehensive analysis of genetic alterations underlying MEC has not been reported. In order to shed light on the genomic landscape of MEC, we performed whole-exome sequencing and copy-number analysis of tumors from 18 patients.

A retrospective review of patients treated for MEC was performed after obtaining approval from the Institutional Review Board of the Johns Hopkins Medical Institutions. Clinical and demographic data including age, gender, tobacco use, primary site, and lymph node involvement were extracted from electronic medical records. Tumors were scored using the Armed Forces Institute of Pathology grading scheme (11).

Formalin-fixed, paraffin-embedded (FFPE) tissues were dissected to achieve a neoplastic cellularity of >60%. DNA was purified from these tumors, as well as matched non-neoplastic tissue adjacent to tumor, using the AllPrep DNA/RNA purification Kit (Qiagen; catalog # 80204) according to the manufacturer's instructions. Extracted DNA was then used to generate libraries suitable for massively parallel sequencing.

Sample library construction, next-generation sequencing (NGS), and bioinformatic analyses of tumor and normal samples were performed at Personal Genome Diagnostics. In brief, genomic DNA from tumor and normal samples was fragmented and used for Illumina TruSeq library construction (Illumina) and captured using the Agilent V4 exome panel as per the manufacturer's instructions. Paired-end sequencing, resulting in 100 bases from each end of the fragments, was performed using a HiSeq 2000 Genome Analyzer (Illumina).

Somatic mutations were identified using VariantDx (12) custom software for identification of mutations in matched tumor and normal samples. Prior to mutation calling, primary processing of sequence data for both tumor and normal samples was performed using Illumina CASAVA software (v1.8), including masking of adapter sequences. Sequence reads were aligned against the human reference genome (version hg18) using ELAND. Candidate somatic mutations, consisting of point mutations and small (<50bp) insertions, and deletions were then identified using VariantDx across the coding exomic regions. VariantDx examines sequence alignments of tumor samples against a matched normal while applying filters to exclude alignment and sequencing artifacts. In brief, an alignment filter was applied to exclude quality failed reads, unpaired reads, and poorly mapped reads in the tumor. A base quality filter was applied to limit inclusion of bases with reported Phred quality scores >30 for the tumor and >20 for the normal (http://www.phrap.com/phred/). A mutation in the tumor was identified as a candidate somatic mutation only when (i) distinct paired reads contained the mutation in the tumor; (ii) the number of distinct paired reads containing a particular mutation in the tumor was at least 10% of read pairs for exome; (iii) the mismatched base was not present in >1% of the reads in the matched normal sample as well as not present in a custom database of common germline variants derived from the Single Nucleotide Polymorphism Database of the National Center for Biotechnology Information (NCBI); and (iv) the position was covered in both the tumor and normal. Mutations arising from misplaced genome alignments, including paralogous sequences, were identified and excluded by searching the reference genome. Candidate somatic mutations were further filtered based on gene annotation to identify those occurring in protein-coding regions. Functional consequences were predicted using snpEff and a custom database of CCDS, RefSeq, and Ensembl annotations using the latest transcript versions available on hg18 from UCSC (https://genome.ucsc.edu/). Predictions were ordered to prefer transcripts with canonical start and stop codons and CCDS or Refseq transcripts over Ensembl when available. Finally, mutations were filtered to exclude intronic and silent changes, while retaining mutations resulting in missense mutations, nonsense mutations, frameshifts, or splice site alterations. A manual visual inspection step was used to further remove artifactual changes. We have optimized our sequencing and bioinformatics approaches in the past so that specificity of mutations is extremely high. This has been extensively validated not only by Sanger sequencing but also by NGS at high depth. A minimum of 95% of the mutations identified using these approaches are bona fide (13, 14). Copy-number alterations were identified by comparing normalized average per-base coverage for a particular gene in a tumor sample to the normalized average per-base coverage in a matched normal sample for that patient (15). Focal amplifications (≥3-fold or six copies) and homozygous deletions were reported.

MECT1-MAML2 translocation was determined by FISH performed on FFPE sections using a commercially available MAML2 dual-color, break-apart probe (Z-2014-200; Zytovision). Prior to hybridization, the slides were deparaffinized using a VP 2000 processor (Abbott Molecular) in which pretreatment with protease I was used. Following deparaffinization, the slides and the MAML2 probe were codenatured at 80°C for 7 minutes and allowed to anneal over night at 37°C in humidified atmosphere. At the end of the incubation, the slides were washed in 2 X SSC/0.3% NP-40 for 2 minutes at 72°C and for 2 minutes at room temperature, with agitation. Traces of detergent were removed with a wash in 2 X SSC at room temperature. The slides were counterstained with DAPI, and a cover slip was applied using Vectashield mounting medium (H-1000; Vector Laboratories, Inc.).

A fluorescence microscope was used to evaluate the probe pattern. Cells with two fusion signals of one orange and one green fluorochrome were scored as normal. Cells with rearrangements for MAML2 gene had one normal fusion signal and one orange and one green signal at a distance from each other.

All statistical analyses were carried out with two-sided tests with statistical significance level set at P value of 0.05. STATA 10 software (Stata Corp) was used for the analyses.

Whole-exome sequencing was performed for 18 tumors with matched normal tissue as described in the Materials and Methods section. The average high quality per base coverage of the normal samples was 101-fold and 174-fold for the tumors, with 90% and 92% of targeted bases represented by at least 10 reads, respectively (Supplementary Table S1).

Using stringent criteria for analyses of these data (16), we identified 774 candidate somatic mutations in 705 genes among the 18 tumors (Supplementary Table S2). The range of mutations per tumor was 3 to 242, with a mean and SD of 43 ± 63.5 mutations per tumor (Table 1).

There were differences in the genetic landscapes of tumors based on histologic grade. Intermediate- and high-grade tumors tended to harbor TP53 mutations (5 of 9), while low-grade tumors did not have any TP53 mutations (0 of 9; P = 0.03, Fisher exact test; Supplementary Table S3). More C:G>G:C substitutions were seen in intermediate- and high-grade tumors compared with low-grade tumors (P = 0.03, Fisher exact test; Supplementary Table S4).

Somatic mutations were identified in at least two tumors in eight different genes (Table 1; Supplementary Table S2). TP53 was the most frequently mutated gene, with mutations found in five tumors. Among the five TP53 mutations, two were frameshift alterations and three were missense mutations (Q331H, G245S, R280T). Tumors harboring TP53 mutations had significantly more mutations overall than tumors without TP53 mutations (104.4 ± 99.6 vs. 17.8 ± 14.2, P = 0.006, Student t test). The only two high-grade tumors without known MECT1-MAML2 had TP53 mutations. One of these patients, who presented with locally advanced disease and later developed distant metastasis, was found to have a truncating TP53 mutation (insertion frameshift).

Three recurrent in-frame deletions were observed in POU6F2; mutations in this gene were found exclusively in low-grade MEC tumors, but the difference was not statistically significant, potentially due to the limited sample size (P = 0.21, Fisher exact test). Other genes with mutations in more than one tumor included IRAK1, MAP3K9, ITGAL, ERBB4, OTOGL, KMT2C, and OBSCN. Of these, MAP3K9 and OTOGL mutations were seen only in intermediate- or high-grade tumors.

Several known oncogenes and tumor-suppressor genes were noted to harbor mutations in a single tumor each. Mutations were observed in the oncogenes ARID1A, CBL, ABL1, AR, EPHA5, FH, INSR, PRKDC, RET, and HRAS and the tumor-suppressor genes PBRM1, SMAD2, SMAD3, FBXW7, and HNF1A.

Copy-number analysis revealed 34 CNVs in nine tumors, of which all but one were focal amplifications. One recurrent CNV occurred in two intermediate-grade tumors (tumor samples 17 and 20) with amplification of 17q21.2. One high-grade tumor (tumor sample 29) had amplification of 19q13.2-q13.3. KLK2 amplification (tumor sample 29) was the only potentially biologically significant gene.

FISH was employed to determine the presence or absence of the MECT1-MAML2 translocation (Supplementary Fig. S1). The translocation was present in 15 of 17 tumors, including all of the low- and intermediate-grade tumors and three of five high-grade tumors; translocation status could not be determined for one patient with intermediate-grade MEC (Table 1).

Whole-exome sequencing of 18 MEC samples revealed that TP53 mutation burden differed based on histologic grade of the tumors. The most frequently mutated gene identified through these analyses was TP53, though TP53 mutations were isolated to intermediate- and high-grade tumors. We also identified mutations in POU6F2, IRAK1, MAK3K9, ITGAL, ERBB4, OTOGL, KMT2C, and OBSCN in more than one tumor. Copy-number analyses demonstrated that most observed CNVs were amplifications. We also confirmed MECT1-MAML2 translocation in 88% of tumors.

Our analyses found that the most commonly mutated gene in MEC is TP53, the most frequently altered gene in human cancer (17). TP53 mutations have been previously reported in 25% to 33% of MEC, although only small numbers of MEC cases were included in those studies, which were not performed in an unbiased fashion (18–20). Of the five tumors with TP53 mutations in our study, three were missense mutations, and two were frameshift alterations. TP53 mutations were identified in only intermediate- or high-grade tumors, consistent with one study that suggested detection of aberrant p53 expression using immunohistochemistry was associated with higher histologic grade. This study also found that aberrant p53 expression was associated with local recurrence (21). Furthermore, we found that the presence of TP53 mutation was significantly associated with higher mutation frequency overall. However, the clinical implications of these findings are hard to assess given the heterogeneity of the patients. Further research is therefore warranted to investigate possible role of the mutation as a prognostic biomarker.

POU6F2 was the second most frequently mutated gene, with mutations found in three low-grade MECs. All three POU6F2 mutations were in-frame deletions at the same location (187Q>-) and were found in low-grade MEC tumors without TP53 mutation. In COSMIC, the same in-frame deletion was seen in 12 specimens from various human cancers, including lung adenocarcinoma, endometrioid carcinoma, melanoma, hemangioblastoma, esophageal squamous cell carcinoma, intestinal adenocarcinoma, and Wilms tumor. Among those, TP53 mutations were seen in only three of 12 specimens. POU6F2 encodes a member of the POU protein family, characterized by the presence of two DNA binding subdomains (a POU-specific domain and a homeodomain) separated by a variable polylinker. The POU family members are transcriptional regulators, many of which are known to control cell type–specific differentiation pathways (22). POU6F2 is involved in the development of the pituitary (23) and kidney (24). LOH in regions containing POU6F2 has been reported in Wilms tumor (25). Further studies are necessary to elucidate the role of POU6F2 in human cancers, including MEC.

The other recurrently mutated genes included IRAK1, MAP3K9, ITGAL, ERBB4, OTOGL, KMT2C, and OBSCN. MAP3K9 (26) and ERBB4 (27) encode for protein kinases that have been frequently implicated in human cancers and may be potential therapeutic targets. IRAK1 encodes IL1 receptor-associated kinase 1, which is a serine/threonine protein kinase that associates with the IL1 receptor upon stimulation. IRAK1 can mediate stimulation of the NF-κB and MAPK pathway, thereby promoting cell proliferation (28). ITGAL encodes the integrin alpha L chain, which combines with the beta 2 chain to form LFA-1, which plays a critical role in cell–cell interactions, cellular adhesion, and cytotoxic T-cell–mediated killing (29, 30). KMT2C encodes for mixed-lineage leukemia 3 (MLL3), which acts as a histone H3 lysine 4 (H3K4)–specific methyltransferase (31). Although mutations in KMT2C have been reported in various solid tumors, its role as a true driver of oncogenesis is debated (32, 33). As MLL3 is a coactivator of p53, its inactivation may result in reduced p53 function and subsequent accumulation of genetic damage (34). The role of OTOGL and OBSCN in MEC tumorigenesis is unclear. Alterations in genes with potential clinical utility included recurrent mutations in ERBB4 and isolated mutations in ARID1A, ABL1, INSR, RET, and HRAS.

Using FISH, we verified the presence of the MECT1-MAML2 translocation in 88% of tumors, slightly higher than the previously reported range of 38% to 81% (35). Due to limited availability of fresh-frozen tumors, we were unable to perform whole-genome sequencing to detect MECT1-MAML2 or other chromosomal rearrangements. Consistent with the literature, all low- or intermediate-grade MEC tumors in this study were positive for the translocation, whereas 60% of high-grade tumors were positive (36). MECT1-MAML2 translocation produces a fusion protein that disrupts Notch signaling through the effects of Maml2, which normally acts as a coactivator of Notch receptors and transactivates Notch target genes (8, 37). By forming a novel Mect1-Maml2 fusion product, Maml2 replaces its Notch ligand binding domain with the CRTC1 promoter and CREB binding domain, which allows the fusion protein to act as a coactivator of the cAMP/CREB signaling pathway (38, 39). It has been shown that ectopic expression of Mect1-Maml2 may be tumorigenic in rat epithelial cells, and inhibition of the fusion protein using RNA interference can suppress growth in fusion-positive MEC cell lines (40). A recent report suggests that the fusion protein may also interact with Myc and activate Myc transcription targets, including genes involved in cell growth, metabolism, survival, and tumorigenesis (41). These findings, along with lack of significant driver oncogene mutations in MEC, strongly suggest that MECT1-MAML2 translocation may be the main oncogenic driver in this tumor type. In our analysis of 18 patients, 2 patients who did not harbor the MECT1-MAML2 translocation had TP53 mutations, including a frameshift insertion and a missense mutation in the DNA binding domain (R280T), which provides an alternative mechanism of tumorigenesis in the absence of the translocation.

In recent years, comprehensive genomic analyses have been performed in a number of salivary gland neoplasms (42). Chromosomal translocations resulting in gene fusions and mutations in known cancer pathways appear to be of particular importance in several tumors of salivary origin. In adenoid cystic carcinoma, for example, the t(6;9) translocation resulting in the MYB-NFIB fusion product is well-described, and two recent NGS studies found alterations in chromatin regulation and Notch signaling pathways (43, 44). The significance of the MECT1-MAML2 translocation and of TP53 mutation demonstrated here in MEC is therefore consistent with findings in other salivary tumors.

The results of this study provide evidence that MEC tumors have few common genetic aberrations other than the MECT1-MAML2 translocation and TP53 and POU6F2 mutations. The MECT1-MAML2 translocation was seen in all low-grade tumors and most intermediate- and high-grade tumors; in two high-grade cases without MECT1-MAML2 translocation, we identified TP53 mutations. The presence of TP53 mutations was significantly associated with higher histologic grade and a higher number of mutations overall. POU6F2 may also represent a driver mutation as three mutations were found in low-grade MEC. Future investigation to elucidate the functional roles of MECT1-MAML2 translocation and the somatic mutations in MEC tumorigenesis would be highly desired.

No potential conflicts of interest were disclosed.

Conception and design: M. Sausen, P.K. Ha, N. Agrawal

Development of methodology: M. Sausen, N. Agrawal

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.A. Bishop, P.K. Ha, N. Agrawal

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Kang, M. Tan, S. Jones, M. Sausen, N. Agrawal

Writing, review, and/or revision of the manuscript: H. Kang, M. Tan, J.A. Bishop, S. Jones, M. Sausen, P.K. Ha, N. Agrawal

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P.K. Ha, N. Agrawal

Study supervision: P.K. Ha, N. Agrawal

We thank our patients for their courage and generosity and R. Yonescu for technical assistance.

This work was supported by NIH/NIDCR grant DE023218 (N. Agrawal) and NIH/NIDCR Specialized Program of Research Excellence grant DE019032 (N. Agrawal).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.