Genomic Landscape Comparison of Cardiac versus Extra-Cardiac Angiosarcomas

Angiosarcomas (ASs) are rare malignant vascular entities that can affect several regions in our body, including the heart. Cardiac ASs comprise 25–40% of cardiac sarcomas and can cause death within months of diagnosis. Thus, our aim was to identify potential differences and/or similarities between cardiac and extra-cardiac ASs to enhance targeted therapies and, consequently, patients’ prognosis. Whole-transcriptome analysis of three cardiac and eleven extra-cardiac non-cutaneous samples was performed to investigate differential gene expression and mutational events between the two groups. The gene signature of cardiac and extra-cardiac non-cutaneous ASs was also compared to that of cutaneous angiosarcomas (n = 9). H/N/K-RAS and TP53 alterations were more recurrent in extra-cardiac ASs, while POTE-gene family overexpression was peculiar to cardiac ASs. Additionally, in vitro functional analyses showed that POTEH upregulation conferred a growth advantage to recipient cells, partly supporting the cardiac AS aggressive phenotype and patients’ scarce survival rate. These features should be considered when investigating alternative treatments.


Introduction
Angiosarcomas (ASs) are malignant soft-tissue tumors characterized by an incidence of 1-2% among all sarcomas and poor prognosis [1][2][3].This subtype arises from endothelial cells of blood or lymphatic vessels and can affect different areas, including the head, neck, breast, bone, and viscera [3].The median age at diagnosis ranges between 52 and 67 years old, and the distribution between men and women is analogous [2,4].
Low-grade angiosarcomas present a solid component along with low-grade cytology and vessels with open lumina, while high-grade angiosarcomas display high mitotic rates and atypical cells [5].Angiosarcomas can also be classified according to their onset.Primary ASs arise de novo, whereas secondary ASs develop following chronic lymphedema or exposure to radiotherapy or vinyl chloride [2,3,6].Other potential risk factors include UV radiation, arteriovenous fistula, and a compromised immune system [6].
Visceral angiosarcomas comprise cardiac sarcomas, which represent 25-40% of cardiac sarcomas [7].Cardiac ASs primarily arise in the right atrium and have a higher incidence in men with a 2/3:1 ratio [7][8][9].Malignant cells can also invade the pericardium and cardiac chambers causing obstruction and, subsequently, heart failure [8].Additionally, this subgroup is characterized by a high proliferation rate and late diagnosis, since symptoms are aspecific and onset is rare.Thus, patients are usually deceased within months of diagnosis [9].Surgical resection alone remains the most effective option, especially when a tumor is strictly localized; however, a combination of surgery, chemotherapy, and radiation is crucial to prolong overall survival.
Considering the rarity and the heterogeneity in their possible site of origin, depicting a well-defined AS mutational profile has been challenging.Therefore, in the present study, we evaluated the molecular differences and similarities between cardiac and extra-cardiac angiosarcomas using next-generation sequencing.

Materials and Methods
From 2005 to 2017, 11 patients diagnosed with primary cardiac sarcomas were identified in our center.This case series comprised several rare subtypes including tree angiosarcomas (27%) [10,11], two of which were suitable for genomic analysis (L359 and L360) (SRA Accession Number: PRJNA896891).
Case L359: A 74-year-old male affected by a right atrial lesion of 7 cm was diagnosed with cardiac MRI and PET-FDG (SUVmax 43.7).He underwent surgery with explorative and diagnostic intent followed by biopsy.The histology was fused, high-grade angiosarcoma (Figure 1).He began one cycle of chemotherapy with doxorubicin and died of the disease after a few weeks.
Low-grade angiosarcomas present a solid component along with low-grade cytology and vessels with open lumina, while high-grade angiosarcomas display high mitotic rates and atypical cells [5].Angiosarcomas can also be classified according to their onset.Primary ASs arise de novo, whereas secondary ASs develop following chronic lymphedema or exposure to radiotherapy or vinyl chloride [2,3,6].Other potential risk factors include UV radiation, arteriovenous fistula, and a compromised immune system [6].
Visceral angiosarcomas comprise cardiac sarcomas, which represent 25-40% of cardiac sarcomas [7].Cardiac ASs primarily arise in the right atrium and have a higher incidence in men with a 2/3:1 ratio [7][8][9].Malignant cells can also invade the pericardium and cardiac chambers causing obstruction and, subsequently, heart failure [8].Additionally, this subgroup is characterized by a high proliferation rate and late diagnosis, since symptoms are aspecific and onset is rare.Thus, patients are usually deceased within months of diagnosis [9].Surgical resection alone remains the most effective option, especially when a tumor is strictly localized; however, a combination of surgery, chemotherapy, and radiation is crucial to prolong overall survival.
Considering the rarity and the heterogeneity in their possible site of origin, depicting a well-defined AS mutational profile has been challenging.Therefore, in the present study, we evaluated the molecular differences and similarities between cardiac and extra-cardiac angiosarcomas using next-generation sequencing.

Materials and Methods
From 2005 to 2017, 11 patients diagnosed with primary cardiac sarcomas were identified in our center.This case series comprised several rare subtypes including tree angiosarcomas (27%) [10,11], two of which were suitable for genomic analysis (L359 and L360) (SRA Accession Number: PRJNA896891).
Case L359: A 74-year-old male affected by a right atrial lesion of 7 cm was diagnosed with cardiac MRI and PET-FDG (SUVmax 43.7).He underwent surgery with explorative and diagnostic intent followed by biopsy.The histology was fused, high-grade angiosarcoma (Figure 1).He began one cycle of chemotherapy with doxorubicin and died of the disease after a few weeks.
Case L360: A 39-year-old male affected by a right atrial lesion of 6 cm was diagnosed with a CT scan, MRI, and PET-FDG (SUVmax 8).He underwent tumor resection, which was not completed due to vena cava infiltration.The histology was angiosarcoma (grade 3) with a high mitotic index (14/10 high-power field-HPF) (Figure 1).He began chemotherapy with taxol and died of the disease.In addition to these two cases, to evaluate the molecular differences and similarities between cardiac and extra-cardiac angiosarcomas, AS samples from the Gene Expression Case L360: A 39-year-old male affected by a right atrial lesion of 6 cm was diagnosed with a CT scan, MRI, and PET-FDG (SUVmax 8).He underwent tumor resection, which was not completed due to vena cava infiltration.The histology was angiosarcoma (grade 3) with a high mitotic index (14/10 high-power field-HPF) (Figure 1).He began chemotherapy with taxol and died of the disease.
In addition to these two cases, to evaluate the molecular differences and similarities between cardiac and extra-cardiac angiosarcomas, AS samples from the Gene Expression Omnibus database (https://www.ncbi.nlm.nih.gov/geo/,accessed on 27 September 2022; accession number: GSE163359) were included in this study-one cardiac and twelve extracardiac non-cutaneous ASs, whose RNA had been extracted from fresh tissue (Table 1) [12].

Coding Transcriptome Sequencing
Formalin-fixed paraffin-embedded (FFPE) slides were manually macrodissected by a pathologist to achieve a tumor tissue enrichment of at least 70%.Total RNA was extracted using a RecoverAll Total Nucleic Acid Isolation Kit (Thermo Fisher Scientific, Waltham, MA, USA), and cDNA libraries were synthesized from 100 ng total RNA using a TruSeq RNA Exome kit (Illumina, San Diego, CA, USA) following manufacturer's instructions.cDNA library synthesis included RNA fragmentation, adapter ligation, amplification, and probe hybridization to select coding exon sequences.Subsequently, libraries were sized with Agilent DNA 7500 chips on the Bioanalyzer 2100 (Agilent Technologies, Taipei, Taiwan) and quantified with a fluorometric assay (Quant-IT Picogreen assay; Life Technologies, Carlsbad, CA, USA).Paired-end libraries were sequenced at 80 bp on a NextSeq500 instrument (Illumina, San Diego, CA, USA).

Bioinformatic Analysis
First, quality control of all the FASTQ files was performed with FastQC [13], and the results across all samples were summarized using MultiQC [14].Two samples (L410 and SRR13260960) were excluded from the analysis due to a high number of overrepresented sequences.Reads of the remaining samples were aligned and mapped on the reference human genome hg38 (No Alts, with decoys) followed by RNA quantification on BaseSpace, an Illumina web tool (http://euc1.sh.basespace.illumina.com/,accessed on 7 October 2022).Small variant calling combined with variant annotation with Nirvana was performed.
Regarding gene alterations, filters were applied to vcf output files to select only stop gain/splice donor/splice acceptor/splice region/frameshift indels/inframe deletion/inframe insertion/initiator codon/ATG loss/missense mutations predicted as deleterious or probably damaging with the SIFT predictor, with a GnomAD frequency <0.01, a QC metric quality >29, a total read depth >14, an alt allele depth >3, and a variant read frequency >0.15.Variants were validated using the Integrative Genomic Viewer (IGVhttps://igv.org/,accessed on 10 December 2022).Lastly, only genes belonging to Tier 1 of the Cancer Gene Census list (https://cancer.sanger.ac.uk/census, accessed on 10 December 2022) were considered for further analysis.Sequencing data were analyzed using the DRAGEN RNA app (version 3.10.4).Our mutational events were also compared to the ones from the Angiosarcoma Project-Count Me In, with a focus on cutaneous ASs [15].These additional data (48 samples from 36 patients) were retrieved from the cBioPortal for Cancer Genomics database (https://www.cbioportal.org,accessed on 1 December 2022), including one cardiac angiosarcoma and nine cutaneous ASs of the head, neck, face, and scalp (HNFS).Further analysis was performed using STAR to map paired reads on the reference human genome hg38 [16]; subsequently, duplicate removal, sorting, and indexing were applied using Samtools [17].Gene expression was quantified and normalized as transcripts per million (TPM) from raw gene counts generated with the Python package HTseq-count [18].TPM were then used to perform a principal component analysis (PCA) using the R package prcomp [19] and a differential gene expression (DGE) analysis applying the Mann-Whitney U test (p-value < 0.05), following batch correction with the sva package [20].Batch correction was necessary since, as previously mentioned, our RNA samples had been extracted from paraffin-embedded tissue, while the GEO samples were derived from fresh tissue.

Transfection of the POTEH Gene Transcript
To investigate the potential role of POTEH overexpression, a plasmid DNA (pcDNA3.1(+)-C-eGFP)containing the whole POTEH transcript under the control of a CMV promoter and upstream of an eGFP-tag sequence was synthesized (GenScript, Rijswijk, NL, USA).The expression plasmid was transformed and amplified in One Shot TOP10 Chemically Competent E. coli cells (Life Technologies, Carlsbad, CA, USA) and then purified with an E.Z.N.A. Endo-Free Plasmid DNA Maxi Kit (Omega Bio-tek, Norcross, GA, USA).HEK293 cells were obtained from CLS Cell Lines Service (Eppelheim, Germany) and cultured in Dulbecco's Modified Eagle Medium high glucose (DMEM, Gibco) enriched with 10% fetal bovine serum (FBS), 1% L-glutamine, and 1% penicillin-streptomycin. Subsequently, the cells were seeded on 6-well plates (0.5 × 10 6 cells/well) in DMEM 10% FBS without antibiotics, and after 24 h, they were transiently transfected with POTEH plasmid or with empty pcDNA3.1 vector using Lipofectamine 2000 (Life Technologies, Carlsbad, CA, USA).Mock-transfected (treated only with Lipofectamine) and pcDNA3.1-transfectedcells were used as controls.To assess transfection efficiency, GFP positivity was quantified using a FACSCanto flow cytometer (Beckton Dickinson, Franklin Lakes, NJ, USA).To measure the POTEH effect on cell viability, an MTT assay was performed 48 and 72 h after seeding transfected cells in complete medium.Total RNA was extracted 72 h after transfection using the RNeasy Mini Kit (Zymo Research, Irvine, CA, USA), and reverse transcribed with the iScript™ Reverse Transcription Supermix (Bio-Rad, Hercules, CA, USA) into cDNA to quantify POTEH overexpression.

Real-Time PCR Analysis
POTEH expression was evaluated in transfected HEK293 cells using quantitative RT-PCR on a Roche (LightCycler480, Roche Diagnostics) with a Luna Universal One-Step RT-qPCR kit (New England BioLabs, Ipswich, MA, USA).GAPDH and ATPS were used as housekeeping genes.GraphPad PRISM Software was used for statistical analysis.The p-value was estimated against pcDNA3.1 using an unpaired t-test (** p < 0.01; *** p < 0.001).

Gene Expression Analysis
First and foremost, the expression profile of three cardiac versus eleven extra-cardiac non-cutaneous ASs was analyzed (Table 1).The unsupervised PCA did not show a distinct cluster of extra-cardiac angiosarcomas along any principal component; however, the three cardiac samples were located close to one another (Figure 2).This suggested the presence of, at least, some common features among cardiac cases and confirmed AS heterogeneity.

Gene Expression Analysis
First and foremost, the expression profile of three cardiac versus eleven extra-cardiac non-cutaneous ASs was analyzed (Table 1).The unsupervised PCA did not show a distinct cluster of extra-cardiac angiosarcomas along any principal component; however, the three cardiac samples were located close to one another (Figure 2).This suggested the presence of, at least, some common features among cardiac cases and confirmed AS heterogeneity.Additionally, a supervised analysis of cardiac versus extra-cardiac angiosarcomas detected 248 differentially expressed genes (p-value < 0.05).Interestingly, several POTEfamily members, including POTEH, POTEG, and POTEM, were upregulated in cardiac ASs (Figure 3).These findings were further validated by the GSEA rank-ordered gene list, where the aforementioned POTE-related genes were identified as specific gene markers of cardiac ASs.Moreover, the epithelial-mesenchymal transition (EMT), myogenesis, gene downregulation in response to UV, and cardiac and striated muscle contraction pathways were upregulated in cardiac samples compared with extra-cardiac ASs (GSEA Hallmarks, Reactome, and KEGG databases) (Figure 4).Additionally, a supervised analysis of cardiac versus extra-cardiac angiosarcomas detected 248 differentially expressed genes (p-value < 0.05).Interestingly, several POTEfamily members, including POTEH, POTEG, and POTEM, were upregulated in cardiac ASs (Figure 3).These findings were further validated by the GSEA rank-ordered gene list, where the aforementioned POTE-related genes were identified as specific gene markers of cardiac ASs.Moreover, the epithelial-mesenchymal transition (EMT), myogenesis, gene downregulation in response to UV, and cardiac and striated muscle contraction pathways were upregulated in cardiac samples compared with extra-cardiac ASs (GSEA Hallmarks, Reactome, and KEGG databases) (Figure 4).

POTEH Overexpression
Due to the significance of POTE-family overexpression in cardiac ASs, we decided to investigate the role of POTE genes and selected POTEH among its family members due to its pronounced upregulation.We transfected HEK293 cells with an expression vector containing the full-length POTEH transcript and an eGFP tag.After 24 h, 49% of POTEHtransfected cells exhibited GFP positivity and RT-PCR effectively demonstrated a significant increase in POTEH mRNA levels in POTEH-transfected cells when compared with pcDNA3.1 controls.Regarding the phenotypic effect, POTEH overexpression significantly increased cell viability, with a 63% and 42% increase with respect to pcDNA3.1-transfectedcells at 48 and 72 h, respectively (Figure 5).significant increase in POTEH mRNA levels in POTEH-transfected cells when compared with pcDNA3.1 controls.Regarding the phenotypic effect, POTEH overexpression significantly increased cell viability, with a 63% and 42% increase with respect to pcDNA3.1transfectedcells at 48 and 72 h, respectively (Figure 5).

Gene Alteration Analysis
First, comparing mutational signatures between cardiac and extra-cardiac non-cutaneous angiosarcomas led to the identification of the following events in ASs of the heart (Table 2).The RAD51B missense mutation (Lys243Arg) is classified as pathogenic on COS-

Gene Alteration Analysis
First, comparing mutational signatures between cardiac and extra-cardiac non-cutaneous angiosarcomas led to the identification of the following events in ASs of the heart (Table 2).The RAD51B missense mutation (Lys243Arg) is classified as pathogenic on COSMIC and probably damaging with a score = 1 on PolyPhen.In contrast, it is considered likely benign and benign on ClinVar and Varsome, respectively.Additionally, PSIP1 presented a deletion involving four nucleotides (c.651_654del), which caused a frameshift.The KDR c.2267-1_2267 insertion shifting the reading frame is defined as likely pathogenic on Varsome; while the JAK2 point mutation (c.2840G>A) is probably damaging according to PolyPhen and pathogenic with a score = 0.96 on COSMIC.
Alterations concerning RNF213 and ATRX were present in both cardiac and extracardiac ASs (Figure 6) [15].Regarding RNF213, two point mutations (Ile3318Val and Thr1705Lys) are considered benign on ClinVar and Varsome.According to PolyPhen, the Ile3318Val substitution is also benign, while the other two alterations (Thr1705Lys and Pro2274Leu) are probably damaging with a score of 0.98 and 0.99, respectively.In L359, ATRX presented an in-frame deletion (c.6792_6794del) involving exon 31, where a whole triplet encoding glutamic acid was depleted.The same gene presented a point mutation (c.4675A>T) in SRR13260959, causing stop-codon gain.ATRX nonsense mutations were also found in two cutaneous samples [15].Lastly, the TSC2 missense mutation (Ile885Val) is of uncertain significance according to ClinVar and Varsome, tolerated on SIFT, and possibly damaging with a score = 0.88 on PolyPhen.The KMT2C point mutation (c.5053G>T) is defined as deleterious on SIFT, probably damaging on PolyPhen, and pathogenic with a score = 0.99 on COSMIC, but also as benign on ClinVar.FANCE harbored a missense variant (c.1095A>C) classified as deleterious on SIFT and as possibly damaging on PolyPhen.Regarding the NBN alteration c.758C>T, there are conflicting interpretations: it is considered deleterious on SIFT, but benign and likely benign on PolyPhen and ClinVar, respectively.Including cutaneous AS, the most frequent events concerned TP53 (n = 11).TP53 missense mutations (Arg248Gln, Arg248Trp, Arg280Gly, Tyr220Cys, Pro250Leu, Ser241Phe, and Glu286Lys) are frequently associated with cancer development (Figure 6).For instance, on COSMIC, the Arg248Gln variant is present in 173/3610 mutated samples (4.8%), while Arg248Trp is in 152/3610 (4.2%).The c.920-1G>A splice acceptor variant and the Trp53*, Gln317*, and Gln165* nonsense mutations are classified as pathogenic on ClinVar.Other remarkable alterations involved the following genes related to tumor onset.NRAS and HRAS point mutations (Gln61Leu) are considered deleterious on SIFT and possibly damaging with a score = 0.86 and 0.58 on PolyPhen, respectively.On COSMIC, Gln61Leu affecting NRAS is observed in 125/1183 (10.6%) mutated samples, while the same variant affecting HRAS is observed in 30/234 (12.8%).
Lastly, the TSC2 missense mutation (Ile885Val) is of uncertain significance according to ClinVar and Varsome, tolerated on SIFT, and possibly damaging with a score = 0.88 on PolyPhen.The KMT2C point mutation (c.5053G>T) is defined as deleterious on SIFT, probably damaging on PolyPhen, and pathogenic with a score = 0.99 on COSMIC, but also as benign on ClinVar.FANCE harbored a missense variant (c.1095A>C) classified as deleterious on SIFT and as possibly damaging on PolyPhen.Regarding the NBN alteration c.758C>T, there are conflicting interpretations: it is considered deleterious on SIFT, but benign and likely benign on PolyPhen and ClinVar, respectively.

Discussion
The PCA findings confirmed AS heterogeneity, which partly depends on the tumor site.Indeed, the results also highlighted common features between cardiac and extracardiac angiosarcomas.
Among AS-mutated genes, TP53 is well-known for its role as a tumor suppressor; however, the role of its alterations has remained unclear in cardiac angiosarcomas thus far.A study by Garcia and colleagues and the whole-exome sequencing analysis by Zhrebker's team did not identify any TP53 mutations in cardiac ASs [22,23].In support of this, TP53 seems to be frequently altered in extra-cardiac ASs, especially in UV-induced head and neck ASs [24][25][26].In contrast, in a patient affected by cardiac AS, immunohistochemical analyses detected high levels of mutated p53 gene products in tumor cells [22].Another work showed the presence of TP53 point mutations mostly in extra-cardiac ASs, but also in two cases of cardiac ASs [27].These results were further supported by the Angiosarcoma Project [15]: TP53 was the most altered gene (29.2%), and mutational events occurred in one cardiac AS and in ten extra-cardiac ASs, five of which were cutaneous.
Combining these five cutaneous samples [15] with our findings, TP53 alterations were identified mainly in extra-cardiac cases (n = 9) and in one cardiac AS.They were missense and truncating variants, which are known for negatively affecting apoptosis, senescence, and DNA repair.Therefore, although our study only included four cardiac ASs, these findings might suggest that TP53 mutations are, in fact, more common in ASs located outside of the heart.
The second most mutated gene was RNF213, which was recently identified as an earlystage lung cancer biomarker [28].RNF213 alterations might lead to carcinogenesis because of its involvement in angiogenesis, vasculogenesis, inflammation, and proliferation [29].In our study and the Angiosarcoma Project [15], this gene was altered in two cardiac ASs and three extra-cardiac ASs.Thus, RNF213 involvement in AS onset does not seem to strongly correlate with tumor localization.
Alterations affecting the RAS pathway are also remarkable events in AS onset as they can alter cell division, which greatly contributes to tumor development.These signatures are far more common in extra-cardiac ASs [25,26]; indeed, 5/39 extra-cardiac AS cases presented H/K/N-RAS mutations [24].NRAS and HRAS were exclusively mutated in three and two extra-cardiac cases, respectively [15], while KRAS point mutations were only detected in two cases of cardiac ASs [22].Accordingly, our results showed deleterious NRAS and HRAS alterations in two extra-cardiac ASs.NRAS and HRAS are proto-oncogenes involved in cell division; thus, their mutations can lead to uncontrolled proliferation, a feature of cancer cells.Moreover, in the Angiosarcoma Project NRAS was mutated in two breast ASs and one cutaneous AS, while HRAS was mutated in two ASs of the breast [15].
Unsurprisingly, other genes associated with angiogenesis (KDR and PSIP1) were altered in our analysis.KDR, which encodes a vascular endothelial growth factor (VEGF) receptor, was found mutated in 7-10% of soft-tissue ASs, mainly in the breast [7,15,30].This was not confirmed in our samples since both breast ASs lacked KDR alterations.Conversely, seven out of eight patients presenting KDR mutations were diagnosed with breast AS in the Angiosarcoma Project [15].Furthermore, this gene was mutated in a few cases of cardiac ASs [23,31], as it occurred in one of our cardiac samples (SRR13260952).KDR and PLCG1, which also contributes to angiogenesis, are mutually exclusive [30].In this regard, the SRR13260952 case presented solely KDR alterations.Mutations in TP53 and KDR are also mutually exclusive [15]; indeed, as previously mentioned, TP53 alterations only concerned extra-cardiac AS.Both exclusivities were further confirmed in the Angiosarcoma Project [15].
When altered, PSIP1 can enhance angiogenesis and prevent apoptosis in cancer cells [32].Additionally, PSIP1 contributes to homology-directed repair (HDR) [33].In our study, it is not the only altered gene involved in DNA repair: RAD51B participates in homologous recombination repair (HRR) of double-strand DNA breaks [34,35]; FANCE is a member of the Fanconi anemia complementation group (FANC) and plays a role in DNA cross-links repair [36]; NBN is an MRE11-RAD50-NBN (MRN) component involved in DNA recombination, telomere maintenance, cell division regulation and double-strand DNA break repair [35,36]; and lastly, KMT2C is a histone methyltransferase that marks sites for transcription and DNA repair [37].Genes involved in DNA repair were altered in many extra-cardiac ASs and in one cardiac AS.
Alterations in ATRX, JAK2, and TSC2 are also noteworthy since these genes contribute to cell growth and proliferation.ATRX is involved in chromatin remodeling, transcription, and telomere maintenance (ALT pathway) [36,38].In our analysis it was mutated in one cardiac (L359) and in one extra-cardiac case (SRR13260959), while it presented nonsense mutations in two cutaneous ASs.This suggests that ATRX alterations might not be subtype-specific biomarkers.JAK2 plays a major role in growth factor signaling and histone modifications [39,40], while TSC2 is a tumor suppressor that inhibits cell growth by downregulating the mTORC1 pathway [41].Consequently, when altered, these genes might cause uncontrolled cell proliferation.
To better distinguish cardiac from extra-cardiac angiosarcomas, we conducted multiple gene expression analyses, whose results will now be explored.
The POTE-gene family comprises 14 homologous genes, including cancer-testis antigens (CTAs), located on several chromosomes.These genes are usually upregulated in prostate and epithelial ovarian cancer (EOC), including high-grade serous carcinoma (HGSC) [42][43][44] but also in breast and lung cancer [45].Their overexpression in ovarian cancer is associated with more advanced stages and poor prognosis [44].Therefore, POTE-related genes might be therapeutically targeted.To validate these observations, we focused on POTEH, the most upregulated POTE family member in cardiac ASs.This gene is located on chromosome 22 and is expressed in normal prostate, ovarian, and testis tissues and also prostate, ovarian (e.g., HGSC), and lung cancer [43,44,46].In our research, we transfected mammalian cells with a plasmid containing its transcript sequence and, as a result, successfully transfected cells presented greater growth levels compared with pcDNA3.1-transfectedsamples.Since POTEH-overexpressing cells can be considered more proliferative and resistant over time, gene upregulation in cardiac angiosarcomas highlights a more aggressive phenotype and could at least partly explain patients' scarce survival rate.
Furthermore, in cardiac sarcomas, we observed an overexpression of genes involved in the epithelial-mesenchymal transition, which is in line with their threatening histotype.In EMT, endothelial cells tend to detach from surrounding cells and, eventually, acquire the phenotype of mesenchymal stem cells, including the ability to migrate and colonize other tissues.
Two common events in angiosarcomas were not detected in our samples.POT1 is usually mutated in ASs, especially in those located in the head and neck [15,26,47,48].As proof, it was mutated in three cutaneous cases from the Angiosarcoma Project [15].Since it is involved in telomere maintenance and apoptosis, POT1 alterations can lead to longer telomeres and, subsequently, to cell immortality.Mutated POT1 can enhance ATR-dependent DNA damage signaling, which induces cell cycle arrest [48].Subsequently, cell senescence can cause progenitor cell depletion and tissue stress.Therefore, cells tend to bypass apoptosis and acquire additional somatic mutations, which characterize tumoraggressive phenotypes.
Another typical event in ASs is MYC amplification, especially in secondary angiosarcomas following breast cancer [3,49].MYC is relevant to cancer development because of its involvement in oncogenic signaling.Unfortunately, the whole transcriptome analysis applied in this study did not allow us to verify this event.
Until now, the medical treatment of advanced or unresectable ASs has been independent of tumor localization and comprises the administration of anthracyclines and/or gemcitabine/Taxotere, as stated by consensus conference guidelines [50].Although our molecular findings did not suggest site-specific variations in the chemotherapy regimen, emerging data regarding cutaneous AS immune sensitivity are intriguing [15,51,52].Indeed, ipilimumab-nivolumab combined therapy has demonstrated a durable response in ASs, especially in cutaneous tumors of the scalp or face [51].Additionally, cutaneous angiosarcomas of the head, neck, face, and scalp appeared to be characterized by high tumor mutational burden (TMB) and signatures corresponding to UV light exposure [15,52].These features suggest that UV damages might greatly contribute to HNFS-AS development.Cutaneous samples from the Angiosarcoma Project presented a greater number of mutational events compared with cardiac and extra-cardiac non-cutaneous ASs, which can be correlated with abundant immune infiltrate [15].Thus, this AS subtype might be more sensitive to checkpoint inhibitors.
Ultimately, mutational events were more homogeneous in HNFS angiosarcomas than in cardiac ASs, even though both areas are localized.Among the cutaneous samples, the proto-oncogene RET, FGFR2, and NOTCH2 involved in cell growth, proliferation, and apoptosis, and POT1 presented elevated mutational rates, highlighting their role as potential therapeutic targets.Consequently, these differences between cutaneous and non-cutaneous angiosarcomas, and also between cardiac and extra-cardiac ASs, should be carefully considered during treatment selection, although high TMB and UV signature do not ensure promising response to immunotherapy [52].

Conclusions
Our molecular results supported AS heterogeneity but also highlighted some distinct features between cardiac and extra-cardiac ASs.TP53 and H/K/NRAS mutations seemed to be predominant causative factors in extra-cardiac AS development.Conversely, POTE alterations were signatures of cardiac ASs.Since our in vitro results showed that POTEH upregulation enhances cell viability, these biomarkers might be relevant to define and, eventually, to predict AS onset in patients.
However, considering the limited sample size as a drawback in our study and more broadly in rare tumor studies, future attempts should be directed toward analyzing more samples to further confirm our findings and potentially identify new alterations.Wholeexome-sequencing analyses should also be performed to investigate gene amplifications and deletions, which might indicate other biomarkers.
From a therapeutic perspective, our results highlighted a higher mutational burden in cutaneous angiosarcomas compared with non-cutaneous ASs.Moreover, since POTE genes are cancer-testis antigens, their overexpression in cardiac ASs might suggest their amenability to immunotherapy.This might pave the way to explore alternative targets and treatments in non-cutaneous ASs, especially in angiosarcomas of the heart.

Figure 4 .
Figure 4. GSEA analysis of enriched pathways in cardiac versus extra-cardiac non-cutaneous angiosarcomas applying the following gene sets: Hallmarks, Reactome, and KEGG.ES: Enrichment Score; NES: Normalized Enrichment Scores; FDR q-val: False Discovery Rate q-value.