Use of Targeted Next-Generation Sequencing to Identify Activating Hot Spot Mutations in Cherry Angiomas

This case series study examines 10 biopsy specimens obtained from cherry angiomas to investigate mutational patterns that are shared by cherry angiomas and malignant neoplasms.

Key Points

Question

Which somatic mutations are found in benign cherry angiomas?

Findings

In a case series of 10 cherry angioma tissue samples, 5 samples were found to have activating hot spot mutations in GNAQ (Q209H, Q209R, and R183G) and GNA11 (Q209H) genes.

Meaning

Cherry angiomas frequently carry specific genetic variants, which are known to be involved in pathways underlying vascular conditions, such as congenital hemangiomas, hepatic small-vessel neoplasms, port-wine stains, and Sturge-Weber syndrome, as well as melanocytic proliferations, such as blue nevi, melanoma associated with blue nevus, and uveal melanoma.

Abstract

Importance

Shared gene variants in benign-malignant process pairs, such as BRAF mutations common to benign nevi and melanoma, are associated with differing phenotypic manifestations. Study of gene mechanisms underlying cherry angioma may uncover previously unknown disease relationships.

Objective

To identify somatic mutations present in cherry angioma specimens by using targeted next-generation sequencing.

Design, Setting, and Participants

In a single-center case series, 10 formalin-fixed, paraffin-embedded cherry angioma specimens from biopsies performed at Massachusetts General Hospital in Boston from July 10, 2016, to January 23, 2018, were obtained and underwent sequencing across a panel of 323 genes most relevant to cancer. Somatic mutations were curated by excluding variants that were presumed to be germline or of low mapping quality.

Main Outcomes and Measures

Identification of somatic mutations associated with cherry angiomas.

Results

In 10 cherry angioma tissue samples originating from 6 female and 4 male patients with a median (range) age of 54 (26-79) years, 5 samples (50%) revealed somatic missense mutations in GNAQ (Q209H, Q209R, and R183G) and GNA11 (Q209H). Individually, these mutational hot spots are known to be involved in entities that include congenital and anastomosing hemangiomas, hepatic small-vessel neoplasms (Q209), port-wine stains, and Sturge-Weber syndrome (R183). Both hot spots are associated with blue nevi, melanoma associated with blue nevus, and uveal melanoma.

Conclusions and Relevance

In this case series study, the high prevalence of 5 known genetic drivers within the benign cherry angioma entity appears to support the context-dependent role of gene alterations in both benign and malignant proliferations from various cellular origins.

Introduction

Many oncogenic mutations occur with high frequency in both benign and malignant proliferations (eg, BRAF V600E mutations in benign acquired melanocytic nevi and cutaneous melanoma¹), suggesting that oncogene-triggered signaling alone is insufficient to induce cancer. Furthermore, mutations in the same gene may yield pathologically distinct conditions, such as GNAQ mutations in uveal melanoma² and vascular anomalies such as Sturge-Weber syndrome.³ Therapeutically targeting oncogenic drivers has been associated with significant response in cancer, but the roles of such genes in benign conditions are less clear.¹ Uncovering mutational patterns in benign counterparts of malignant processes may yield fresh insights into genomic mechanisms underlying their shared biology. The clinical profile of cherry angiomas (CAs)—mature endothelial cell proliferations⁴—is similar to that of other age-related benign entities, such as nevi, lentigines, and seborrheic keratoses. The genetic landscape of CAs has not been thoroughly described in the literature. Other than polymerase chain reaction–based evidence that supports a possible role for human herpesvirus 8–related genetic material in eruptive CAs,⁵ little is known about the somatic mutations that underlie this cutaneous lesion. We thus sought to use next-generation sequencing to examine CA genetic alterations among a panel of 323 cancer-relevant genes.

Methods

Ten formalin-fixed, paraffin-embedded CA samples from biopsies taken between July 10, 2016, and January 23, 2018, from unrelated individuals (Table) were obtained from the archival tissue collection at the Massachusetts General Physicians Organization Dermatopathology Associates in Newton. Samples were subjected in 2 batches (batch A: 4 samples, patients 1-4, samples obtained on December 6, 2017, from biopsies performed from July 10, 2017, to November 13, 2017; batch B: 6 samples, patients 5-10, samples obtained on February 13, 2018, from biopsies performed from November 17, 2017, to January 23, 2018) to the sequencing pipeline covering a panel of 323 cancer-relevant genes (eMethods and eTable 1 in the Supplement). The dates from obtaining the first tissue banked samples to the receipt of the final sequencing data were December 6, 2017 to May 2, 2018. Of 234 total mutations, we excluded 167 likely germline variants based on the presence of a known single-nucleotide polymorphism annotation or high allele balance; 31 based on the presence of intronic mutations; and 17 BCR and NOTCH2 mutations based on low mapping quality and nonspecific neighboring mutations, indicating an unreliable region (eFigure 1 in the Supplement). Mutations were annotated with a functional impact score, the average of conservation and specificity scores, calculated using evolutionary conservation patterns. Because the conservation and specificity scores are both negative logarithmic functions, mathematically, the functional impact score can range from negative infinity to infinity. Biologically, scores lie within the range –4 and 6, with most scores falling between –2 and 4: higher scores indicate greater likelihood of a functional mutation or driver mutation.⁶ Proportions of endothelial cell nuclei to total endothelial and keratinocyte nuclei were manually estimated on hematoxylin-eosin–stained slides (eFigure 2 in the Supplement). Additional details are provided in the eMethods in the Supplement.

Table. Summary of Key Clinical, Mutational, and Tissue Composition Data in 10 Cherry Angioma Samples.

Patient No./Sex/Age	Location	Mutation	Allele Balance, %a	Endothelial Cell Nuclei, No. (%)b
Batch A
1/F/35	Left side of frontal scalp	None	NA	NA
2/M/26	Right side of inferior chest	GNAQ Q209H	6.4	648 (31.1)
3/F/32	Right side of neck	GNAQ R183G	5.4	614 (25.6)
4/F/47	Left thigh	None	NA	NA
Batch B
5/F/55	Right lower axilla	GNA11 Q209H	8.6	667 (28.5)
6/M/71	Right forearm	GNAQ Q209R	6.2	838 (36.5)
7/M/79	Left dorsal forearm	None	NA	NA
8/M/53	Left lower back	None	NA	NA
9/F/57	Left side of back	GNAQ Q209H	10.9	806 (27.3)
10/F/60	Right thigh	NOTCH1 C552Y	5.8	1356 (23.3)

Abbreviation: NA, not applicable.

^aThe mean (SEM) allele balance was 7.2% (0.9%).

^bThe mean (SEM) number of endothelial cell nuclei was 821.5 (113.0) and percentage of endothelial cell nuclei was 28.7% (1.9%).

The Partners HealthCare Institutional Review Board determined that electronic medical record review and specimen collection were exempt from institutional review board approval and from obtaining written informed consent from patients. R statistical software, version 3.5.0 (R Foundation for Statistical Computing) was used for all statistical analyses. Significance was set at a P < .05; significance testing was 2-sided.

Results

Ten samples from 6 women and 4 men with a median age of 54 years (range, 26-79 years) (Table) revealed 19 somatic variants after excluding putative germline variants, low-quality calls, and potential artifacts (eTable 2 in the Supplement). Cherry angiomas in samples 1, 4, and 7 carried no mutations among the interrogated gene panel.

Only missense mutations in GNAQ (HGNC 4390), GNA11 (HGNC 4379), and NOTCH1 (HGNC 7881) had high mutational functional impact scores (a score ≥3.5)⁶ (Figure 1 and Figure 2). GNAQ variants were the most frequent (40%), specifically the GNAQ Q209H mutation (samples 2 and 9), Q209R (sample 6), and R183G (sample 3). The GNA11 Q209H missense mutation was identified in sample 5, and the NOTCH1 C552Y missense variant in sample 10. The incidence of GNAQ and GNA11 mutations was 50% in both sequencing batches. Additional mutations occurring frequently in the samples (20%) were NF1 (HGNC 7765) nonsense variants (Q282 and E1694 in sample 2 and R440 in sample 3) and STAG2 (HGNC 11355) splice region variants (samples 2 and 8) and had uncertain functional significance. The mean (SEM) allele balance of GNAQ, GNA11, and NOTCH1 mutations was 7.2% (0.9%), and endothelial cell nuclei comprised a mean (SEM) of 28.7% (1.9%) of total (endothelial and keratinocyte) nuclei (Table). Assuming mutation heterozygosity, the estimated tumor cellularity based on allelic frequency was 15% to 20%.

Figure 1. Somatic Mutations Identified Among Cherry Angioma Samples.

Six of 10 cherry angioma samples carried high–functional impact score (FIS) missense mutations in GNAQ, GNA11, and NOTCH1. Mutations with “—” are not missense amino acid substitutions, but nonsense, splice site, and so on; thus, the FIS cannot be calculated for this type of mutation. An explanation of the FIS is provided in the Methods section. The FIS scores in red indicate high-FIS mutations; in blue, medium-FIS mutations; and in black, neutral-FIS mutations. The horizontal line separates high-FIS mutations from the rest. NMD indicates nonsense mediated decay.

Figure 2. Alignment Tracings for GNAQ, GNA11, and NOTCH1 Mutations.

Missense somatic mutations in GNAQ, GNA11, and NOTCH1 were identified in regions of high mapping quality. bp Indicates base pair.

Forty-two percent of base pair changes were C>T transitions at dipyrimidine (CT, TC, and TT) sites (eFigure 3 in the Supplement). Structural variants, including deletions, duplications, and fold-back inversions, ranged between 166 and 490 per sample. Structural variant number did not differ between GNAQ/GNA11-mutant and wild-type samples and was significantly higher among samples from female patients (mean [SEM] 401 [37.7]) than in samples from male patients (207 [18.1]; P = .002) (eFigure 4 in the Supplement).

Discussion

We describe what is, to our knowledge, the first systematic interrogation of variants in a common vascular tumor, the CA. Our identification of GNAQ Q209, GNAQ R183, and GNA11 Q209 mutations in 50% of samples highlights a likely role of GNAQ and GNA11 variants in CA development. The GNAQ and GNA11, closely related large GTPases, mediate downstream signaling from G-protein–coupled receptors. Histologically, CAs are characterized by dilated vessels rather than dense cell infiltrates.⁴ Thus, it is conceivable that GNAQ and GNA11 mutations may lead to aberrations in vessel formation rather than endothelial cell proliferation, which is consistent with these mutations being implicated in various types of vasculopathy, such as Sturge-Weber syndrome.³

The mutations identified in this study suggest a possible genetic association of CAs with other entities. Mutations at both Q209 and R183 hot spots in GNAQ and GNA11 have been identified in uveal melanoma,^2,7 blue nevi,² and melanoma associated with blue nevus.² Mosaic mutations in GNAQ R183, GNAQ Q209, and GNA11 R183 have also been associated with phakomatosis pigmentovascularis, which comprises vascular and dermal pigmentary anomalies.⁸ The Q209 position specifically is a known hot spot for single⁹ and multifocal¹⁰ congenital hemangioma, hepatic small-vessel neoplasms,¹¹ and anastomosing hemangioma.¹² In addition, the R183 position (but not Q209) is a hot spot for nonsyndromic port-wine birthmarks,³ Sturge-Weber syndrome,³ and endothelial capillary malformations.¹³ Our data suggest involvement of both R183 and Q209 hot spots in CA development. R183 mutations have been proposed to activate the p38 mitogen-activated protein kinase (MAPK) pathway and Q209 mutations to activate p38 MAPK, c-Jun N-terminal kinase (JNK), and extracellular-signal-regulated kinase (ERK) pathways. Further study of these genetic overlaps may help unravel why heterogeneity of these hot spots is observed in some conditions, such as uveal melanoma and CA, but not in others, such as Sturge-Weber syndrome.

We identified a high–functional impact score somatic NOTCH1 C552Y mutation not previously described in the literature. Notch intercellular signaling is known to play a crucial role in vascular development and physiology, with notch signaling pathway defects associated with inherited vascular disease.¹⁴ However, given the presence of NOTCH1 in only 1 of the 10 samples in our study, this is a putative CA-associated finding requiring further supportive evidence (ie, similar mutations in more samples, further functional validation). We identified 2 samples (20%) with inactivating NF1 nonsense mutations. Cherry angiomas occur more frequently in persons with neurofibromatosis type 1 than in the general population,¹⁵ suggesting a possible association between NF1 inactivation and CA pathogenesis.

Limitations

Our study has several important limitations. First, next-generation sequencing indiscriminately analyzes DNA from all cell types within specimens; thus, variants specific to angioma cells cannot be separated from those in other cells. To address this limitation, we estimated the angioma-to–total cell percentage as 28.7% (Table), which is likely an upper bound given that stromal (nonkeratinocytic) nuclei could not be quantified. Thus, somatic angioma-specific heterozygous mutations would be expected to occur with an allele balance of at most 14.4% (28.7% divided by 2). The observed mean allele balance (7.2%) is near the expected allele balance, which supports, but does not prove, an endothelial cell origin for the observed mutations. Second, the sample number was limited given that benign CAs are readily recognized clinically and thus rarely sampled for biopsy. Therefore, these results should be viewed as pilot analyses. More samples could also guard against batch effects. We did address this potential artifact by performing sequencing in distinct cohorts. The GNAQ/GNA11 mutation incidence was 50% in each of the separate cohorts, supporting interbatch reliability. Future study with comprehensive sequencing of additional samples and prospective biopsy of several lesions on a single patient may help to strengthen the findings.

Conclusions

We suggest a possible role for GNAQ- and GNA11-mediated signaling in CA pathogenesis, a finding consistent with the genetic underpinnings of vascular anomalies. We observed both R183 and Q209 hot spot variants among CAs, suggesting that CAs are possibly on a shared genetic spectrum with vascular entities, such as congenital and anastomosing hemangiomas, capillary malformations, port-wine birthmarks, and Sturge-Weber syndrome, and melanocytic growths, such as blue nevi, melanoma associated with blue nevus, and uveal melanoma.

Supplement.

eFigure 1. Exclusion of Suspected Germline and Unreliable Variants

eFigure 2. Tissue Slides Used for Visual Estimation of Cellular Composition

eFigure 3. Single Nucleotide Variation Analysis

eFigure 4. Structural Variant Analysis

eMethods. Tissue Preparation, Sequencing, and Bioinformatics Pipeline.

eTable 1. 323-Gene Panel Subjected to Targeted Sequencing

eTable 2. Somatic Mutations Identified by Targeted NGS of Cherry Angioma Samples