Parathyroid Carcinoma: Update on Pathogenesis and Therapy

: Parathyroid carcinoma (PC) is a very rare endocrine cancer with aggressive behavior, a high metastatic potential, and a poor prognosis. Surgical resection of affected gland(s) and other involved structures is the elective therapy. Pre-operative and intra-operative differential diagnosis with benign parathyroid adenoma remains a challenge. The lack of a clear pre-operative diagnosis does not allow one, in many cases, to choose the correct surgical approach to malignant PC, increasing persistence, the recurrence rate, and the risk of metastases. An initial wrong diagnosis of parathyroid adenoma, with a minimally invasive parathyroidectomy, is associated with over 50% occurrence of metastases after surgery. Genetic testing could help in identifying patients at risk of congenital PC (i.e., CDC73 gene) and in driving the choice of neck surgery extension. Targeted effective treatments, other than surgery, for advanced and metastatic PC are needed. The pathogenesis of malignant parathyroid carcinogenesis is still largely unknown. In the last few years, advanced molecular techniques allowed researchers to identify various genetic abnormalities and epigenetic features characterizing PC, which could be crucial for selecting molecular targets and developing novel targeted therapeutic agents. We reviewed current ﬁndings in PC genetics, epigenetics, and proteomics and state-of-the-art therapies.


Introduction
Parathyroid carcinoma (PC) is an extremely rare endocrine malignancy of the parathyroid glands, representing about 1% of all parathyroid tumors and one of the rarest causes of primary hyperparathyroidism (PHPT), and generally presenting more severe symptomatic hypercalcemia than its benign counterparts (hyperplasia and adenoma), with marked skeletal and renal complications, including osteoporosis, fragility fracture, osteitis fibrosa cystica, and nephrolithiasis [1]. This tumor has aggressive behavior and high metastatic potential. Patients with PC show a poor prognosis, with an overall survival rate of 85% and 49%, respectively, at 5 and 10 years after diagnosis, due to commonly unmanageable, severe, and drug-refractory hypercalcemia [2]. Tumor stage, adjacent tissue invasion, and malignant local and distant metastases, at the time of first diagnosis, still represent critical prognostic factors.
The majority of PC occur as sporadic cancer; only occasionally, it is part of congenital syndromic and non-syndromic endocrine diseases, such as hyperparathyroidismjaw tumor syndrome (HPT-JT) or isolated familial hyperparathyroidism (FIPH), and, in extremely rare cases, multiple endocrine neoplasia type 1 and type 2A (MEN1 and MEN2A, respectively).
The diagnosis of PC is still a challenge. Some biochemical, clinical, and radiological features may help in discriminating PC by benign adenoma, in some cases. Currently, the diagnosis of malignant carcinoma principally derives from a combination of operative findings by surgeons, such as vascular invasion, invasion of the recurrent nerve or esophageal wall, thick adherent capsule, the presence of lymph node metastases, a tumor mass over 2 g, and the presence of four or more histological characteristics of malignancy in a tumor sample assessed by pathologists, such as intra-tumoral bands, signs of necrosis, the presence of small cells with a large nucleus (macronuclei) and a reduced cytoplasm/nucleus ratio, diffuse cellular atypia, the presence of more than 6 mitoses per 10 high-power field, signs of capsular disruption, and invasion of adjacent tissue [3]. However, even with the presence of such features, a clear diagnosis of PC remains difficult, and it is operator-dependent. The loss of nuclear parafibromin immunostaining in the tumor specimen may further confirm the intra-operative and histological suspicion of PC.
Recently, a high expression of cancer-derived immunoglobulin G (CIgG) was found in PC samples, being significantly higher in PC patients than in patients with parathyroid adenoma (PA) or hyperplasia (p < 0.001) and borderline significantly higher in recurrent or metastatic lesions of PC compared with the primary tumor (p = 0.055) [4], suggesting the possibility to use this parameter as a possible adjunctive biomarker of PC diagnosis and for the prediction of disease relapse in clinical practice.
PC specimens were also shown to have a remarkable loss of nuclear staining of the Yes-associated protein 1 (YAP1) [5], an increased cytoplasmic expression of the Filamin A (FLNA) protein [6], and a reduced number of cells positive for the expression of the T-box transcription factor 1 (TBX1) [7], with respect to PA and/or normal parathyroids. However, these molecular features need to be confirmed before they could be used as markers to distinguish PC from benign PA.
An additional challenge is to distinguish between the recurrence of PC after surgery and the possible occurrence of post-operative parathyromatosis, an extremely rare cause of recurrent primary hyperparathyroidism, consisting of the formation of several nodules of hyperfunctioning parathyroid tissue in the neck and mediastinum due to the incidental seeding of benign parathyroid cells in neck soft tissue after surgical procedures. The parathyromatosis lesion(s) do not reflect the histological and genetic characteristics of the primary PC, and they are usually located in different anatomic sites with respect to PC recurrence and/or metastases.
Surgical resection of affected gland(s) and other involved tissue is still the elective therapy for PC, but it has limited efficacy in the case of advanced and metastatic cancer. Therefore, the identification of genetic and epigenetic drivers of parathyroid carcinogenesis and tumor malignant aggressiveness, as well as of pathways specifically deregulated in PC tumor cells, is fundamental to selecting potential molecular targets for the development of targeted treatments.
In this review, current findings in PC genetics, epigenetics, and proteomics and stateof-the-art therapies were reviewed.

Pathogenesis of Parathyroid Carcinoma
The etiology of PC remains still to be fully elucidated, and it is not yet known whether PC arises de novo or as a malignant progression of parathyroid hyperplasia and adenoma. The rarity of the disease and the lack of studies have made it impossible to solve this issue so far. Interestingly, in patients with chronic kidney disease-derived secondary hyperparathyroidism, a frequent cause of parathyroid hyperplasia/adenoma, including those with end-stage renal disease who commonly experience multiple parathyroid adenomas, the malignant transformation of the disease has never been reported.
The contemporary presence, in some patients, of PC, PA, and/or atypical parathyroid adenoma (aPA) and also the occurrence of different parathyroid phenotypes in members of the same family, without any genotype-phenotype correlation, in the inherited forms of PC make the elucidation of the specific etiological bases of PC very complex.
A germline genetic background has been established for the congenital PC forms, often in association with somatic additional genetic abnormalities in tumor cells. The occurrence of various somatic genetic alterations has been described in subsets of sporadic PC cases, presenting a variable spectrum among patients.
Recently, specific epigenetic alterations, including the hypermethylation of gene promoters, modification of the chromatin structure, and altered expression of microRNAs (miRNAs) and long non-coding RNAs (lncRNAs), have been associated with PC and are suspected to have a role in tumor suppression inactivation and in the induction/progression of carcinogenesis.

Genetics of Parathyroid Carcinoma
Biallelic loss-of-function mutations of the CDC73 tumor suppressor gene are the most common genetic defect of sporadic and congenital PC forms but are found in less than 5% of PAs, suggesting that loss of tumor suppressor activity of the encoded protein can confer the mutated cell aggressive growth potential and malignant behavior and that the identification of CDC73 mutations in tumor cells can be a molecular marker to distinguish between malignant and benign parathyroid neoplasms.
Somatic inactivation or the loss of at least one CDC73 allele is found in 9-100% of sporadic PC, depending on the analyzed series [1], and homozygote CDC73 inactivation/loss is reported in 9-70% of cases [8]. In the inherited PC forms, heterozygote germline inactivating mutations of CDC73 are detected in 50-75% of HPT-JT pedigrees and in about 8% of FIPH families [9]. In these hereditary PCs, CDC73 mutations consist of a "first germline hit", inherited by the affected parent and present in all cells, and a somatic inactivation/loss of the second allele in the tumor-originating parathyroid cell (in most cases due to a large deletion at the 1q31.2 locus).
The presence of a germline mutation of the CDC73 gene not only predisposed to the development of malignant PC but was associated with the risk of disease recurrence after parathyroidectomy [10].
Germline and somatic mutations of the CDC73 gene identified, to date, in the sporadic form of PC and in the context of HPT-JT syndrome and FIPH are summarized in Table 1,  Table 2, and Table 3, respectively.    CDC73 encodes a nuclear ubiquitously expressed, evolutionarily conserved protein, parafibromin, an essential component of Polymerase-associated Factor 1 (PAF1), involved in the regulation of the elongation of gene transcripts and mRNA 3 -end processing, gene expression, chromosome stability, cell cycle regulation, apoptosis, and nuclear transduction of the Wnt canonical signaling. Truncating mutations (nonsense and frameshift variants), which represent more than 80% of CDC73 pathogenic variants in PCs, prevents parafibromin from the nuclear localization and, thus, from the exertion of its biological functions, conferring to the mutated cell a growth advantage and, thus, being responsible for PC carcinogenesis. Mutations reported only in FIHP-associated atypical parathyroid adenoma and/or benign parathyroid adenoma were not included in the table.
In addition to gene mutations, epigenetic silencing of the CDC73 gene, by hypermethylation of the 5 UTR promoter region, has been described as a somatic event in PC [70].
The absence of nuclear parafibromin immunohistochemical staining in tumor specimens is a common hallmark in PC, with a sensitivity and a specificity of 75-90% and over 90%, respectively, which could be used as a marker, together with other histological features, to distinguish PC and PA [8], and it appears to also be a promising predictor of PC prognosis, in particular as an indicator of a higher risk of cancer recurrence and metastasis, and of mortality [71].

CCND1 Gene
The CCND1 gene encodes cyclin D1, a positive regulator of the cell cycle, through the promotion of G1-S phase transition via the activation of the cyclin-dependent kinases CDK4/CDK6. Cyclin D1 has also a role in chromosome stability, whose deregulation could be involved in PC tumorigenesis. The over-expression of cyclin D1 has been reported in up to 90% of PCs [72].
A somatic heterozygote 11p15-q13 pericentromeric inversion, positioning the 5regulatory element of the parathyroid hormone (PTH) gene upstream to the CCND1 gene, was found in two unrelated cases of sporadic PA, but not in PC. The transgenic mouse model mimicking this chromosomal rearrangement at the Ccnd1 locus developed hyperplastic parathyroid glands, chronic PHPT with hypercalcemia, and bone abnormalities but not PC [73]. These data, together with the fact that cyclin D1 up-regulation is reported in 20-40% of PAs [9], seem to indicate that the over-expression of cyclin D1 alone may not be sufficient to drive malignant carcinogenesis, and the presence of other concurrent genetic or epigenetic alterations is suspected to synergize in leading to the malignant phenotype.
A whole-genome study identified an alteration of the CCND1 locus in 29% of the analyzed sporadic PC cases, showing that it is due to the amplification of the gene [13]. Zhao et al. [73] showed that the amplification of the CCND1 gene was more prevalent in PCs than in adenomas (71% vs. 21%), suggesting that the gain of copy number of the CCND1 gene is the principal mechanism of cyclin D1 over-expression in PCs. Interestingly, 80% of CCND1-amplified PCs and cases with somatic inactivation/loss of the CDC73 gene were mutually exclusive [73], suggesting that CCND1 amplification and CDC73 inactivation represent two alternative mechanisms leading to the over-expression of cyclin D1 and the subsequent enhancement of cell growth.

PRUNE2 Gene
In 2015, germline and somatic mutations of the PRUNE2 tumor suppressor gene were identified, for the first time, as associated with PC [12]. The whole-exome sequencing analysis of 8 unrelated sporadic PCs identified three mutations in exon 8 of the PRUNE2 gene (18% of the analyzed cases) but not in all 40 screened PAs [12]. The selective sequencing of exon 8 in a validation group of 13 PC samples found two additional heterozygote somatic missense mutations in 2 PC cases (15.4%) [12].
A heterozygote germline missense mutation (p.Val452Met) was identified in one PC from the discovery group, associated with somatic LOH of the PRUNE2 locus on chromosome 9, while two heterozygote somatic missense mutations (p.Ser450Asn and p.Gly455Asp) were found in two PCs from the validation group, with all these three variants in the absence of a CDC73 mutation [12]. In the discovery group, two heterozygote somatic nonsense mutations (p.Glu474X and p.Glu537X) were identified in a single PC sample, presumably inactivating both the alleles of the PRUNE2 gene, and in the presence of a somatic heterozygote mutation of the CDC73 gene but no LOH of the gene. These two nonsense mutations and the p.Val452Met variant were found, respectively, at the somatic level in one PC sample and in the germline in one PC case by another whole-exome sequencing study [13], suggesting that they could be recurrent mutations involved in the development of sporadic PCs. The same study [13] also identified one novel germline missense variant (p.Glu2570Ala) in one PC case.
The PRUNE2 gene encodes the homonym protein that exerts various biological functions, including the suppression of Ras homolog family member A activity, which results in reduced stress fiber formation and the suppression of oncogenic cellular transformation. Therefore, the inactivation/loss of this protein could exert a role in driving parathyroid malignant carcinogenesis and tumor progression, presumably by regulating the differentiation, survival, and aggressiveness of the tumor cells and their capacity to metastasize.

MEN1 Gene
Germline loss-of-function mutations of the MEN1 tumor suppressor gene are responsible for the development of the MEN1 syndrome, a rare inherited multiple endocrine tumor syndrome in which PHPT, caused by multiple parathyroid hyperplasia and/or adenoma, is the main and, in the majority of cases, the first occurring clinical manifestation. The occurrence of PC in MEN1 patients is extremely rare, and only less than 15 cases have been reported to date [9].
Two genomic profiling studies on sporadic advanced PC samples identified somatic alterations of the MEN1 gene in 3/23 (13.0%) [19] and 6/16 (37.5%) [48] of the screened cases. A somatic mutation of the MEN1 gene was found in the trachea metastasis from a sporadic PC, but not in the primary tumor [59].

Genes of the PI3K/AKT/mTOR Signaling Pathway
The PI3K/AKT/mTOR signaling is a critical cell signal transduction pathway involved in the regulation of cell metabolism, growth, survival, and senescence, as well as in angiogenesis, whose constitutive activation is a common event in various human malignancies, including endocrine neoplasms, being responsible for increased cell proliferation, tumor cell metastatic spread, and tumor angiogenesis.
One gene fusion and one copy number loss of the PI3KCA gene were found in PC samples by Hu et al. [47] through whole-genome sequencing. The same study found somatic mutations in the PI3K/AKT/mTOR pathway in 64.3% (9/14) of the analyzed PC samples from the discovery cohort samples, showing that they were positively correlated with cancer recurrence/metastasis [47]; the overall prevalence of somatic mutations in the PI3K/AKT/mTOR pathway was 78.3% (18/23) in the discovery and the extension cohorts [47].
Interestingly, Zhang et al. [75] demonstrated that inactivating CDC73 mutations could confer parathyroid cancer cells a selective growth advantage by exerting, like mTOR, an inhibitory activity on the eukaryotic translation initiation factor 4E-binding protein (EIF4EBP) gene, encoding the homonym translation repressor, a known effector of mTOR signaling. Under physiological conditions and in its unphosphorylated form, EIF4EBP binds to the eucaryotic translation initiation protein (EIF4E) to inhibit mRNA-cap-dependent initiation of the translation of specific proteins, critical for cell proliferation and survival. The activation of the mTOR pathway leads the protein kinase mTOR to phosphorylate EIF4EBP and reduce its affinity and binding to EIF4E, inducing the translation of target proteins and promoting cell growth. Zhang et al. [75] showed that the promoter region of the EIF4EBP gene was a conserved target of parafibromin and that levels of the EIF4EBP protein were reduced in peripheral blood cells of patients with a CDC73 germline heterozygote inactivating mutation. A haploinsufficiency of parafibromin, due to the presence of a heterozygote CDC73 mutation, resulted in a reduction in/loss of EIF4EBP gene expression and the subsequent loss of its inhibition on the pro-translation activity of EIF4E, as it happens in the presence of a constitutively active mTOR kinase.

Genes of the Wnt Signaling Pathways
The Wnt signaling pathways are pleiotropic signal transduction pathways that regulate cell growth and survival and apoptosis through the modulation of gene expression. Constitutive or aberrant activation of Wnt signaling has been reported in human cancers. PC tumor samples were shown to have an increased cytoplasmatic accumulation of the non-phosphorylated active form of beta-catenin, the intracellular effector of the Wnt signal transduction in the Wnt canonical pathway, compared to the adjacent normal parathyroid tissues [76], suggesting that aberrant activation of canonical Wnt signaling could be a driver of parathyroid carcinogenesis.
Somatic inactivating genetic variants have been found in PC tissues in two genes involved in the regulation of Wnt signaling, the APC (p.Thr297Ile, p.Glu1284Lys, p.Ala1793Gly) [13,28] and the RNF43 (p.Gly659fs) [13] genes, respectively encoding the APC protein, a main inhibitor of beta-catenin activation, and the RNF43 transmembrane ubiquitin-protein ligase, which negatively regulates both the canonical and the noncanonical Wnt pathways by selectively ubiquitinating frizzled receptors and promoting their proteasomal degradation. The inactivation of APC or RNF43 leads to aberrant activation of Wnt signaling and the subsequent transcription of target genes, such as the two regulators of the cell cycle S-phase, c-MYC and CCND1, resulting in the promotion of cell proliferation.
A heterozygote missense mutation, p.Arg380Gln, of uncertain clinical significance in the WNT1 gene, encoding the homonym WNT1 ligand, the activator of the canonical Wnt pathway, was found in a PC sample also bearing a pathogenic mutation of the CDC73 gene [28].
Strangely, a whole-exome study identified an inactivating missense mutation (p.Gln72X) of the CTNNB1 oncogene, encoding beta-catenin 1, in two different formalin-fixed paraffinembedded samples from the same PC patient of the primary tumor and the lung metastasis [2,59].

Other Genes
In the last decade, whole-exome and whole-genome sequencing studies identified somatic variants/alterations in various genes in PC samples [2,13,28,48,59], but the pathogenic role of all these genes in PC carcinogenesis is still unknown.
Interestingly, the contemporary somatic complete deletion of the CDKN2A and CDKN2B genes, lying adjacent at the 9p21.3 locus, was described in two unrelated PC patients [48]. These two tumor suppressor genes encode two inhibitors of the cyclin-dependent kinases (CDKs), the Cyclin-dependent kinase inhibitor 2A (p16INK4A) and the Cyclin-dependent kinase inhibitor 2B (p15INK4B), which induce cell cycle arrest at G1 or G2 phases. The loss of these two cell cycle inhibitors could confer tumor cells an advantage in proliferation. The CDKN2B gene was published by Agarwal et al. as potentially implicated in hyperparathyroidism [77]. A possible role of the loss of tumor suppressor activity of CDKN2A and CDKN2B genes in parathyroid tumorigenesis was also suggested by the finding of promoter hypermethylation of these two genes in PC samples [78].
Two recurrent variants (−146C>T and −124C>T) in the promoter region of the TERT gene, encoding an enzyme with reverse transcriptase activity involved in the maintenance of telomer ends, were found at the somatic level, respectively, in two and three unrelated PC samples [48]. Interestingly, these two mutations were previously identified in 70% of melanoma [79], suggesting they could impair telomerase activity and have a shared role in oncogenesis in different tissues.
A recent gene expression study [80], profiling 740 genes known to be involved in main cancer-progression-related processes, was performed in metastatic and non-metastatic PCs, revealing two specific expression signatures of 87 (44 down-regulated and 43 up-regulated) and 103 (38 down-regulated and 65 up-regulated) genes, respectively characterizing nonmetastatic PCs and metastatic PCs, compared to PAs. A specific sub-panel of significantly up-regulated (ANGPTL4, BMP7, CD24, FGFR1, MMP9, and SOX2) and down-regulated (ERBB3, FBP1, RAB25, and TBX1) genes was identified in metastatic PC, needing further investigation to better define the possible role of these deregulated genes in the malignant progression and metastatic potential of PC.

DNA Methylation
The hypermethylation of the CpG islands in the promoter regions of tumor suppressor genes, resulting in gene expression repression, was reported in various human malignancies, contributing to cancer development by activating cell proliferation.
Interestingly, in PC samples, hypermethylation was reported in promoters of various genes involved in the negative regulation of the Wnt pathway: (1) SFRP1, SFRP2, and SFRP4 [78], encoding three antagonist decoy receptors of Wnt ligands; (2) APC [81], a tumor suppressor encoding a component of the beta-catenin destruction complex; (3) RASSF1A [81,82], a tumor suppressor whose silencing leads to the accumulation of active beta-catenin and activation of Wnt-signaling-regulated gene transcription; and (4) HIC1 [83], encoding a transcription repressor that directly binds beta-catenin and the TCF4 transcription factor, preventing them from activating Wnt-induced TCF-mediated gene transcription. These findings seem to confirm the importance of constitutive/aberrant Wnt pathway activation in the pathogenesis of PC. All these gene promoters, except HIC1, were also hypermethylated in PA but not in normal parathyroid tissue. APC, RASSF1A, and HIC1 were hypermethylated in 100% of analyzed PCs [81][82][83]. Conversely, the hypermethylation of the CTNNB1 gene, encoding the beta-catenin, has only been reported in PAs to date [81].
The hypermethylation of promoters of the GATA4 and the PYCARD genes was specifically found in PC [78]. The GATA4 gene encodes a zinc finger transcription factor that regulates gene expression by binding to the GATA motifs in the promoters of many genes; the role of its silencing in PC tumorigenesis remains to be elucidated. The PYCARD gene encodes an adaptor protein that is involved in the regulation of apoptosis via the activation of caspases; the silencing of the PYCARD gene may confer parathyroid tumor cells the ability to escape apoptosis during carcinogenesis.

Chromatin Modification
Reversible modification of the chromatin structure regulates the accessibility of gene promoters to transcription factors, regulating gene expression, and it is due to posttranscriptional modifications of histones, such as methylation/demethylation, acetylation/deacetylation, and phosphorylation/dephosphorylation. Alterations in enzymes and regulatory molecules responsible for modifications of histones have been reported in cancers.
Parafibromin, which is commonly lost in both inherited and sporadic PCs, is a key regulator of histone modifications. Under normal conditions, parafibromin promotes, through direct interaction with the SUV39H1 histone methyltransferase complex, histone 3 methylation on lysine residues 4, 36, and 79 (H3K4me, H3K36me, and H3K79me, respectively), resulting in transcriptionally active chromatin, and on lysine residue 9 (H3K9me), which leads to transcriptional repression of the CCND1 gene and cell cycle arrest. Moreover, by interacting with the RNF20/RNF40 ubiquitin ligase complex, parafibromin promotes the monoubiquitination of histone 2B on lysine residue 120, a modification that positively regulates RNA elongation during gene transcription.
Amplification of the EZH2 gene at the 7q36.1 locus, encoding histone 3 lysine 27 methyltransferase, was found to be a common event in PC cases (about 60% of cases). A significant over-expression of EZH2 mRNA and protein was reported in malignant PC, with respect to both PAs and hyperplastic parathyroids [84], resulting in increased trimethylation of histone 3 on lysine residue 27 (H3K27me3) commonly associated with the repression of transcription of target genes. Among putative target genes of EZH2-mediated H3K27me3, there is HIC1, a tumor suppressor normally involved in the control of parathyroid cell proliferation, whose silencing through promoter hypermethylation was reported in one study in all five analyzed sporadic PCs [84]. Moreover, the EZH2 protein mediates the transcription repression of some Wnt signaling inhibitors, such as AXIN2, a component of the betacatenin destruction complex, confirming once again that the activation of the Wnt signaling pathway appears to be an important pro-oncogenic event in parathyroid malignancies.

microRNAs
The deregulation of the expression of miRNAs, short non-coding RNA molecules that negatively regulate gene expression at the post-transcriptional level, has been reported in human malignancies, with both over-expressed miRNAs acting as oncogenes (oncomiR) or silenced/down-regulated miRNAs normally exerting tumor suppressor activity.
In agreement with the general behavior of miRNAs in cancer, PCs showed a globally lower expression of miRNAs than the normal parathyroid tissue [85]. miRNAs whose deregulated expression was specifically reported in PC compared to healthy parathyroid and benign PAs, their known target genes, and the possible role of the deregulated miRNA in PC tumorigenesis are summarized in Table 5. Table 5. miRNAs whose expression was found to be deregulated in parathyroid carcinoma.

miRNA [Reference]
Known Target   Encoding the CRK adaptor protein that binds to several tyrosine-phosphorylated proteins and is involved in regulating cell adhesion, spreading, and migration

EGFL7
Encoding the epidermal growth factor-like protein 7 (EGFL7) involved in cell migration and angiogenesis

HOXA9
Encoding the Homeobox A9 (HOXA9) transcription factor involved in embryonic development, morphogenesis, and cell differentiation. Positive regulator of the eukaryotic translation initiation factor 4E (EIF4E) that promotes protein translation initiation

IRS1
Encoding the insulin receptor substrate 1 (IRS1), an intracellular signaling adaptor protein that integrates and coordinates numerous biologically key extracellular signals within the cell KRAS (proto-oncogene) Encoding the KRAS GTPase that activates the RAS/MAPK pathway inducing cell proliferation.

PI3K
Encoding the Phosphatidylinositol 3-Kinase (PI3K) that activates the PI3K-Akt pathway in response to a variety of growth factors, leading to cell proliferation

SLC7A5 (LAT1)
Encoding the SLC7A5 membrane protein of the intracellular organelles that acts as a neutral amino acid transporter

SOX2
Encoding the SRY-box transcription factor 2 (SOX2), involved in the regulation of embryonic development and in the determination of cell fate, required for pluripotent stem-cell maintenance

VEGF
Encoding the vascular endothelial growth factor (VEGF), a growth factor that induces proliferation and migration of vascular endothelial cells, essential for both physiological and pathological angiogenesis miR-296-5p [85] HGS Encoding the Hepatocyte growth factor-regulated Tyrosine kinase substrate that negatively regulates signal transduction mediated by cytokines and growth factors by inducing internalization and degradation of membrane receptors by lysosomes.
In human PC samples, miR-296-5p down-regulation resulted in increased HGS mRNA levels and in an immunostaining-detected dramatic over-expression of HSG [85].
Up-regulated miR-517C positively correlated in vivo with serum calcium and PTH levels, and with parathyroid tumor weight [87].

LATS2 (tumor suppressor)
Encoding the large tumor suppressor kinase 2 (LATS2) that plays a critical role in centrosome duplication, maintenance of mitotic fidelity, and genomic stability, negatively regulates G1/S transition by down-regulating cyclin E/CDK2 kinase activity, and acts as negative regulator of YAP1 in the Hippo signaling pathway restricting cell proliferation and promoting apoptosis Encoding the p27 kyp1 cyclin-dependent kinase inhibitor that inhibits the complex CDK2/CDK4, acting as a negative regulator of cell cycle progression at G1 In human PC samples, the up-regulation of miR-222 resulted in an almost complete loss of expression and nuclear localization of the p27 kyp1 protein [85].

miR-503 [85] CCND1
Encoding the cyclin D1, a positive regulator of cell cycle that promotes the G1 to S phase transition through activation of CDK4 and CDK6 In human PC samples with up-regulation of miR-503, cyclin D1 displayed a heterogeneous immunoreactivity [85].
Out of these miRNAs, miR-296 was almost universally down-regulated in cancer compared to matched normal tissues, and it was often lost in metastatic lesions. The transcriptional profile of tumor cells with the down-regulation of miR-296 is enriched in multiple key regulators of cell motility, invasion, and metastasis [89], indicating this miRNA as a negative regulator of cell migration and suggesting its down-regulation as a marker of poor prognosis. miR-296 was also found to be down-regulated in lung metastasis from a PC [87].
Additionally, to these miRNAs specifically characterizing PC, miR-139-3p showed a similar expression in PC and PA, being significantly down-regulated in both parathyroid tumor types, with respect to the normal parathyroid tissue [85].
In addition to their intracellular localization, the presence of miRNAs has been reported in many biological fluids. miRNAs circulating in the blood (circulating miRNAs, c-miRNAs), both as free molecules and within exosomes, have recently attracted attention as possible high-quality biomarkers of human cancer [90]. Recently, Krupinova et al. [16] analyzed the expression of 754 c-miRNAs in the serum samples from 13 patients with PC and 11 patients with PA, finding 17 c-miRNAs to be significantly down-regulated in the PC group, including 1 miRNA whose down-regulated expression had been previously reported in PC samples [86], miR-126-5p. Among the 17 down-regulated c-miRNAs, miR-342-3p was the only one meeting the Benjamini-Hochberg criteria for multiple comparisons, and it appeared to be a potential diagnostic biomarker for the differential diagnosis of PC. Previous studies showed miR-342-3p as a miRNA with a potential tumor suppressor activity, crucial for the induction of apoptosis and inhibition of cell growth, tumorigenesis, and tumor cell invasion/migration. Currently, the role of the down-regulation of this miRNA in PC is still not known.

Long Non-Coding RNAs
LncRNAs are non-coding transcripts longer than 200 nucleotides, acting as epigenetic regulators of gene expression, mainly in a tissue-specific manner. Accumulating evidence indicates that the deregulated expression of lncRNAs may play a role in cancer biology [91].
Profiling analyses of lncRNAs in PC showed the down-regulation of lncRNA GLIS2-AS1 [92] and the up-regulation of lncRNA PVT1 [92] and lncRNA BC200 [93] in PC compared to PA, being able to distinguish between the two types of tumors.
The over-expression of lncRNA BC200 was reported in a broad spectrum of tumor cells, presumably being responsible for cell viability, invasion, and migration [94]. Interestingly, the over-expression of lncRNA BC200 not only distinguished between PCs and PAs, but this lncRNA was also significantly over-expressed in PC cases bearing CDC73 mutations compared to wild-type carcinomas, in association with a more aggressive clinical phenotype and characterized by higher levels of PTH and calcium [93], suggesting the existence of a pro-oncogenic circuit involving lncRNA deregulation and loss of the tumor suppressor CDC73 that may have a role in parathyroid carcinogenesis and prognosis as an epigenetic modulator. Recently, Morotti et al. [95] quantified the expression of circulating lncRNA BC200 in the serum of 27 patients with PA and 4 patients with PC, showing significantly higher levels in the PC group. Moreover, a measurement of circulating lncRNA BC200 expression in the pre-and post-operative serum samples from 3 PC patients showed that the levels of this lncRNA in serum were significantly reduced after parathyroidectomy. These data suggested the over-expression of lncRNA BC200 as a hallmark of PC and the biochemical dosage of this molecule in the serum as a non-invasive suitable biomarker for initial PC diagnosis, tumor staging, and, possibly, clinical and therapeutic follow-up.
LncRNA PVT1 is encoded by the human PVT1 oncogene, located at 8q24.21, and its transcription was shown to be regulated by the proto-oncogene c-MYC [96], which was demonstrated to promote lncRNA PVT1 accumulation in cancers. Increased copy number and over-expression of lncRNA PVT1 have been associated with various types of human malignancies [9]. Interestingly, lncRNA PVT1 was demonstrated to promote cancer cell proliferation in non-small cell lung cancer through the epigenetic regulation of large tumor suppressor kinase 2 (LATS2) by directly binding histone 3 lysine 27 methyltransferase EZH2, which was previously reported to be over-expressed by gene amplification in PC, acting as a pro-oncogenic factor in parathyroid carcinogenesis [84].
To date, the biological function of lncRNA GLIS2-AS1 is still unknown, and no reports in cancer or other human diseases are available.

Proteomics of PC
To date, only one proteomic study has been performed in PC, comparing the global protein expression profile between PC and PA coexisting in the same patient [97]. Validation was performed additionally in 10 PCs and 10 PAs. A subset of 33 differentially expressed proteins (10 down-regulated and 23 up-regulated) was found in the discovery PC sample and confirmed in the 10 additional PCs, including proteins mainly involved in cell signaling, cell metabolism, and ubiquitin-mediated protein degradation. In particular, PC was characterized, compared to PA, by the over-expression of Ubiquitin C-terminal hydrolase L1 (UCH-L1), a thiol protease involved both in the processing of ubiquitin precursors and ubiquitinated proteins, which recognizes and hydrolyzes proteins bound at the C-terminal glycine of ubiquitin. UCH-L1 was previously demonstrated to act as an oncoprotein in breast cancer, promoting/increasing tumor invasion and metastasis through the activation of the TGFβ [98], the MAPK/Erk [99], and the Akt [100] signaling pathways and enhancing multidrug resistance of cancer cells [99,101].
Other proteins that were specifically over-expressed in PC were 1) Chloride intercellular channel protein 1 (CLIC1), a nuclear chloride channel that is implicated in the activation of the cAMP-dependent protein kinase A that has been associated with cell growth and differentiation, cell-matrix adhesion, cell migration, and metastasis [102]; 2) Malate dehydrogenase 1 (MDH1), a cytosolic enzyme that catalyzes the NAD/NADH-dependent, reversible oxidation of malate to oxaloacetate in many metabolic pathways, and it has been shown to be over-expressed in cancer and associated with poor prognosis [103]; and 3) superoxide Dismutase (SOD2), a mitochondrial protein that destroys superoxide anion radicals derived from oxidative phosphorylation, which showed both tumor suppressive and promoting functions and whose over-expression was identified as a major protein change in fibroblasts associated with ovarian cancer cells [104].
Further studies on larger series of PCs and PAs, preferentially from the same patients to reduce the influence of exogenous confounding factors, are needed to confirm these results.

Surgery
Surgery is the first-line therapy for PC. High pre-operative clinical suspicions of PC and/or the intra-operative identification of malignant features are fundamental in driving surgical management and are key to have a better curative chance. Pre-operative biopsy is contraindicated because of the high risk of tumor capsule rupture and tumor cell dissemination.
The surgical approach to PC consists of the en bloc resection of primary cancer with negative margins, usually associated with the excision of ipsilateral thyroid lobes and adjacent involved structures. Compartmental level VI lymph node dissection is indicated in case of metastatic tumor. The use of an intra-operative measurement of PTH is useful to confirm the removal of all hyperfunctioning parathyroid glands.
However, despite every surgical effort, disease recurrence occurs in over 50% of PC cases, and no specific therapeutic strategy is available to treat such patients.
Prophylactic parathyroidectomy to prevent PC in patients with germline CDC73 mutations is not indicated, since not all carriers develop a malignant tumor. Bilateral exploration of the neck and the identification of all parathyroids glands, with resection of all affected gland(s), are the suggested approaches.

Adjuvant Therapies
Post-operative external beam radiation therapy is rarely used because PC is generally considered a "radioresistant" tumor. Data on the efficacy of radiation therapy on PC are contrasting and appear to be related to the tumor stage and the presence of local and distant invasion [105,106]. The treatment is prevalently restricted to palliative therapy of advanced metastatic disease.
Systemic cytotoxic chemotherapy is used even less frequently than radiotherapy and used only in a few extremely restricted cases, since it showed very limited benefits in inoperable PC patients and patients in whom surgery was ineffective, generally resulting in the inability to control tumor progression and functional tumor burden [106,107].

Therapies for the Control of Calcium Homeostasis
These pharmacological therapies are aimed to control PC-derived severe hypercalcemia before surgical intervention or in patients with inoperable and recurrent tumors or to mitigate the effect of post-operative permanent hypoparathyroidism after tumor resection.
The first-line treatment of hypercalcemia consists of intravenous hydration, diuretics, and anti-resorption drugs, such as bisphosphonates (typically intravenous administration of pamidronate or zoledronic acid) or denosumab, to inhibit osteoclast activity and reduce serum calcium levels.
In PC patients who do not respond to bisphosphonates or for whom the denosumab effect weakens over time, calcimimetic drugs, selective agonists acting on the calciumsensing receptor on the membrane of parathyroid cells, can be used. Cinacalcet, a longacting calcimimetic molecule, showed to be effective in reducing both serum calcium and PTH concentrations in about two-thirds of metastatic and inoperable PCs [108,109].
Post-operative chronic hypoparathyroidism and hypocalcemia require life-long therapy with calcium and active vitamin D.

Emerging Therapies
The administration of some medical therapies targeting molecular pathways showing alterations in a percentage of PC cases, which are currently approved for the treatment of different types of cancer but not PC, showed promising results in some case reports. These therapies need to be further evaluated and confirmed in larger PC case series.

Inhibitors of the PI3K/AKT/mTOR Pathway
Since somatic gene mutations that constitutively activate the PI3K/AKT/mTOR pathway were found in about up to 20% of PC cases [8], the use of PI3K/AKT/mTOR inhibitors could be effective in a subset of PC patients.
Kutahyalioglu et al. [59] treated a patient with a somatic mutation of the TSC1 gene, a known regulator of the PI3K/AKT/mTOR pathway, and recurrent PC with liver metastasis by using a combination of everolimus (an mTOR inhibitor) and vandetanib (an antiangiogenic drug). Treatment allowed a stable disease, with better control of serum calcium levels, while on drugs.
Given the extreme rarity of PC, no randomized controlled clinical trial with PI3K/AKT/ mTOR inhibitors has been conducted on this parathyroid cancer. "Basket trials", evaluating treatments in multiple cancer types sharing common alterations of the PI3K/AKT/mTOR genes, may help in overcoming the problem of patient rarity.

Inhibitors of Angiogenesis
The administration of drugs blocking angiogenesis showed promising results in some case reports in patients with PC.
Mutations in the KDR gene, encoding vascular endothelial growth factor receptor 2 (VEGFR2), were found in 13% of PCs [48]. After the identification of an activating somatic missense mutation (p.Thr688Lys) of the KDR gene in a patient with PC, Kang et al. [48] treated, sequentially, this patient with three inhibitors of VEGFR2, cabozantinib (a potent inhibitor of multiple receptor tyrosine kinases, including MET, RET, AXL, KIT, TIE2, and FLT3), ramucirumab (a monoclonal antibody against VEGFR2), and regorafenib (pleiotropic inhibitor of VEGFR1, VEGFR2, and TIE2). Cabozantinib treatment showed to reduce intact PTH (iPTH) and the size of the paratracheal lymph node over 3 months of treatment but induced various side effects, such as fatigue, hypertension, diarrhea, and epistaxis. After cabozantinib discontinuation, iPTH rose again. Treatment with ramucirumab failed to reduce iPTH levels. Treatment with regorafenib was well tolerated and effective in dropping the iPTH levels. Data from this case report indicate the benefit of performing genetic profiling of PC tumor samples to allow the choice of targeted therapy.
Based on the previously reported good response to the antiangiogenic sunitinib treatment in renal cell carcinoma patients with KDM5C mutations, Kutahyalioglu et al. [59] treated with sorafenib (an antiangiogenic multi-targeted tyrosine kinase inhibitor with a target profile similar to sunitinib) a patient with a somatic mutation of the KDM5C gene and persistent PC with lung metastases, which was not responsive to cinacalcet and intermittent doses of bisphosphonates. The treatment allowed good control of calcemia, even after the discontinuation of cinacalcet, for three years. Then, both PTH and calcium levels started to rise again with the progression of pulmonary metastases. The switch to a more potent antiangiogenic molecule, lenvatinib, granted a radiographically stable disease and good control of calcemia, without any calcimimetic administration, during the 20 months of therapy follow-up reported in the study.
Rozhinskaya et al. [32] positively treated, off-label, a CDC73-mutated woman with PC and multiple lung metastases by using sorafenib, an antiangiogenic multi-kinase inhibitor that blocks the activity of VEGFRs, resulting in the normalization of PTH and calcemia, the prevention of tumor progression, and a significant reduction in the size of lung metastases.
Very recently, Makino et al. [18] showed the effectiveness of combined therapy with sorafenib, denosumab, and evocalcet (a calcium-sensing receptor agonist) to treat refractory hypercalcemia in a CDC73-mutated patient with recurrent PC due to multiple lung metastases.

Conclusions
PC is an extremely rare malignant tumor of the parathyroid with aggressive behavior and a poor prognosis. The diagnosis of PC is still a challenge, and the exact pathogenesis is still largely unknown. In the last few years, advanced molecular techniques, such as whole-exome sequencing, allowed researchers to identify specific genetic abnormalities and epigenetic features characterizing PC and distinguishing PC from benign PA. The identification and functional characterization of molecular drivers and deregulated pathways in PC carcinogenesis are crucial to the development of molecular therapy and novel targeted therapeutic agents to be used in advanced and metastatic inoperable tumors and/or in relapsing PC. Moreover, the identification of personal genetic alterations in the course of clinical care of PC patients could lead to tailored individualized treatments, consisting of rationally matched targeted agents, or immunotherapies, based on the genetic profile of each tumor.
The germinal and somatic loss/inactivation of the tumor suppressor parafibromin and the aberrant activation of the Wnt canonical pathway are two common hallmarks of PC, being suitable targets for tailored therapy in PC patients.
Inhibitors of the PI3K/AKT/mTOR pathway and antiangiogenic drugs showed promising positive results in selected PC case reports, both in the control of hypercalcemia and in the stabilization of cancer.
Unfortunately, the extreme rarity of PC cases and the variable spectrum of genetic and epigenetic alterations among affected patients prevented, to date, the performance of randomized controlled clinical trials to test targeted therapies. The institution of worldwide trials on relatively numerous PC patients is, thus, needed to prove the efficacy of geneticsand epigenetics-based treatments in reducing tumor growth and aggressiveness, and/or in controlling PC-associated refractory hypercalcemia.