Identification of novel GNAS mutations in intramuscular myxoma using next-generation sequencing with single-molecule tagged molecular inversion probes

Intramuscular myxoma (IM) is a benign soft tissue neoplasm that belongs to the group of myxoid tumors characterized by a marked abundance of extracellular myxoid matrix. These tumors share several histological features, and depending on their clinical presentation and place of origin, can be subdivided into intramuscular, superficial-cutaneous, odontogenic and juxta-articular myxoma [1, 2]. These myxomas all represent distinct entities with different characteristic gene lesions involved in their pathogenesis. Therefore, gene mutation analysis can be very helpful in differential diagnosis to support the histopathology of these tumors [3, 4].

IM is characterized by bland spindle- and/or stellate-shaped cells embedded in a hypovascular, abundant myxoid stroma. The nuclei are small showing no or minimal nuclear atypia. Often areas with increased cellularity can be observed and when hypercellular areas predominate it is designated as cellular myxoma [2, 5, 6], which can easily be confused with low-grade fibromyxoid sarcoma or low-grade myxofibrosarcoma, especially in very small biopsies. IM is a somatic mosaic disorder generally occurring as a sporadic solitary neoplasm, although it can be part of Mazabraud’s syndrome characterized by a combination of polyostotic fibrous dysplasia with multiple IM’s [7, 8]. Mazabraud’s syndrome and the closely related McCune-Albright syndrome, which is associated with fibrous dysplasia, café au lait macules and endocrine disorders, are caused by activating missense mutations in codon 201 of the GNAS gene [8‐12].

GNAS encodes the stimulatory G-alpha subunit of the heterotrimeric G-protein complex, which regulates activation of adenylyl cyclase that converts adenosine triphosphate (ATP) into cyclic adenosine monophosphate (cAMP). Overproduction of second messenger cAMP and activation of downstream signaling pathways has been observed in cells harboring GNAS mutations [13, 14]. In 2000, Okamoto et al. first described somatic post-zygotic GNAS mutations in IM with and without fibrous dysplasia [8]. Thereafter, three more studies showed that GNAS lesions occur frequently in sporadic IM, which were detected in 36–61% of the cases, and exclusively involved c.601C > T (p.R201C) and c.602G > A (p.R201H) mutations [8, 15, 16]. On the other hand, GNAS mutations are absent in low-grade myxofibrosarcoma, which can be useful in the differential diagnosis with (cellular) IM [17, 18]. Notably, juxta-articular myxoma and cardiac myxoma also lack GNAS driver mutations [4, 16].

A complicating factor for mutation detection in IM is the mosaicism of GNAS mutations combined with hypocellularity of the tumor, where low concentrations of genomic DNA are isolated from these tissue specimens, especially in the case of biopsy material. In the past decades, several techniques for GNAS mutation detection have been developed and used [8, 19‐21]. In 2009, Delaney et al. tested 28 IM’s for GNAS mutations by using conventional PCR followed by mutation-specific restriction enzyme digestion (PCR-MSRED) and COLD-PCR/MSRED and showed that COLD-PCR/MSRED was more sensitive than the conventional PCR (61% vs. 29% mutations) [15]. Thus, this tumor type may benefit from the development of more robust and sensitive techniques for mutation detection, such as next generation sequencing (NGS). Recently, our molecular diagnostic laboratory has developed a novel NGS-based approach employing single-molecule molecular inversion probes (smMIP) that combines multiplex analysis with single-molecule tagging, also named Unique Molecule Identifiers (UMI) [22, 23]. By using this method, duplicate reads can be identified and merged into a single consensus, reducing false-positive calls originated during PCR and sequencing and allowing a technical sensitivity of 1% mutant allele. In addition, the actual number of sequenced genomic DNA (gDNA) molecules can be determined, which is especially relevant when analyzing limited amounts of gDNA. Furthermore, the strand-specific amplifications allows the distinction between genuine C > T and G > A mutations from deamination artifacts frequently observed when sequencing gDNA in older formalin-fixed paraffin-embedded (FFPE) tissue specimens. [22, 23].

Intramuscular myxoma (IM) mostly occurs sporadically in the skeletal muscle of the thigh. These lesions affect mainly middle-aged adults, women more often than man [1, 24]. The prevailing view is that driver mutations of this neoplasm are exclusively located in codon 201 of the GNAS gene, encoding the stimulatory G-protein alpha subunit that activates the enzyme adenylate cyclase. Due to the low cellularity and somatic mosaicism in most of these lesions, mutation detection can be quite challenging and the presence of a mutation can be easily missed.

In our study we used two different techniques (TaqMan and smMIP assay) to compare the detection sensitivity of GNAS mutations in these lesions. In our series, 23 out of 34 sporadic IM cases (68%) showed a GNAS mutation, 16 out of 29 samples (55%) in the TaqMan assay and 16 out of 28 samples (57%) in the smMIP assay of which 23 samples were successfully analyzed with both techniques showing GNAS mutations in 12 out of 23 (52%) samples. The test-specific detection rate was 55% with the TaqMan assay and 57% for the smMIP approach. The VAF for the TaqMan assay was determined at > 5% in this study and the required input was only 10 ng gDNA. The VAF for smMIP was set at > 1% and a minimum of 3 mutant gDNA molecules, and a coverage of 20 gDNA molecules. This demonstrates that both tests are sensitive methods and useful for molecular diagnostics of tumor samples harboring mutations with a low mutant allele frequency.

In comparison, Walther et al found GNAS mutations in 37% (23/63) of IMs with direct Sanger sequencing and Delaney et al detected mutations in 61% (17/28) using COLD-PCR/MSRED [15, 16]. However, the smMIP technique, because of the whole exon sequencing nature of this test, allowed detection of four additional mutations that previously have not been described in IM. By using smMIP we identified one c.680A > G mutation in exon 9, and three novel mutations in exon 8, one mutation at position c.601, namely c.601C > A, and two mutations at position c.602, which included c.602G > C and c.602G > T. The c.601C > A mutation has previously been reported in fibrous dysplasia, while the c.602G > C and c.602G > T mutations were only reported in sporadic endocrine tumors so far [9, 10, 19, 25]. These mutations were not detected by TaqMan, since this assay was designed to report only the two classical hotspot mutations c.601C > T and c.602G > T.

The smMIP approach allows the distinction between genuine C > T and G > A mutations from deamination artifacts frequently observed when sequencing gDNA from FFPE tissue specimens [22]. All cases harboring a C > T or G > A mutation in GNAS, mutant reads originating from both DNA strands were observed, showing that these represent genuine mutations. Since the TaqMan approach does not allow this distinction, deamination artifact could potentially cause false positive results. In our study, all hotspot mutations detected with the Taqman assay were confirmed with the smMIP technique, indicating no false-positive results with Taqman. Even samples with a VAF of around 5% could be detected by both TaqMan assay and smMIP. The four samples with a hotspot mutation that were detected by TaqMan, but did not yield a reliable smMIP assay result, all had a mutated VAF of approximately 10–30% as judged by Taqman assay, and were therefore interpreted as true mutations.

Significant benefits of the TaqMan assay include low cost and short turn-around time (≤2 working days). A limitation of Taqman is that within one assay only one or two hotspot mutations can be detected. For smMIP analysis the turn-around time in our laboratory is ≤7 working days. A large initial investment was needed and high numbers of samples are required for parallel analyses to have a cost-efficient test. Because multiple genes can be tested at once with the smMIP assay, large amounts of samples are relatively easy obtained in routine clinical setting with the current demand of molecular diagnostics [22]. Because of the sensitive characteristics of the smMIP technique and its accuracy of mutation detection on FFPE material as well as the broader coverage of the GNAS gene, this technique to our opinion is preferable.

The most important differential diagnoses of IM, especially the cellular variant, are low-grade fibromyxoid sarcoma and low-grade myxofibrosarcoma. In biopsy material the distinction can be challenging and in those cases molecular diagnostics can be beneficial. A specific immunohistochemical and molecular signature is well known for low-grade fibromyxoid sarcomas with expression of MUC4 and the presence of FUS/EWSR1-CREB 3 L2/1 fusions making a distinction from IM easily possible [17, 18]. In contrast, for low-grade myxofibrosarcoma, specific immunohistochemical or molecular characteristics are lacking. Sensitive molecular tests, like smMIP and TaqMan assays for GNAS mutation analysis, might be very helpful in assessing the diagnosis, which has therapeutic consequences when considering malignancy [21]. Nevertheless there are also cases in which no GNAS mutation could be detected, suggesting that there are still other aberrations to be identified in IM.

Panagopoulos et al recently found abnormal karyotypes in 21 out of 68 cases, with nine cases showing nonrandom involvement of chromosome 8 (which harbors the GNAS gene) with seven cases showing trisomy 8, one with a deletion and one with a translocation. Only one case in their series showed a c.601C > T GNAS mutation [26]. Thus, chromosomal aberrations could be an alternative explanation for at least a subset of the non-mutated cases.

The smMIP-NGS cancer hotspot panel that was employed to check for GNAS mutations, also contained smMIPs that covered mutational hotspots in the genes AKT1, BRAF, CTNNB1, EGFR, ERBB2, GNA11, GNAQ, H3F3A, H3F3B, HRAS, IDH1, IDH2, JAK2, KRAS, MPL, MYD88, NRAS, PDGFRA and PIK3CA. In none of the 32 samples that could be reliably analyzed by smMIPs, additional mutations were detected in these regions. Thus, GNAS mutations represent a unique driver mutation for this benign tumor type.

In conclusion, both TaqMan and smMIP assay are comparably sensitive molecular methods with valuable applicability in diagnostic pathology for IM. Furthermore, due to a broader coverage of the GNAS gene by the smMIP approach, four novel IM-associated missense mutations of GNAS could be identified (17% of all mutated samples), which previously have only been reported in McCune-Albright syndrome and sporadic endocrine tumors.