Pericytic tumors have been reported to contain other vascular- or perivascular-type tumor components in certain cases. Studies have reported glomus tumors having an overgrowth of vascular smooth muscle cells (glomangiomyoma),24 a hemangioma component (glomangioma or glomuvenous malformation),25 or hemangiopericytomatous vascular channels (glomangiopericytoma).26 Myopericytomas have been reported to be combined with components of angioleiomyoma,2 myofibroma,27 or glomus tumors.26 Although such combined pericytic tumors are generally further classified based on their predominant histologic patterns,2,27 the diagnosis tends to be arbitrary, particularly for tumors with almost equal proportions of different pericytic tumor components or for those with intermediate morphology. However, distinct molecular genetic findings such as PDGFRB and NOTCH3 mutations have been reported as associated with pericytic tumors. Studies have identified PDGFRB mutations as germline mutations in familial infantile myofibromatoses3–5 and somatic mutations in solitary infantile or adult myofibromas.6 Agaimy et al. examined six myopericytomas and nine angioleiomyomas and demonstrated all to be negative for PDGFRB mutations, thus concluding that among all pericytic tumors, PDGFRB mutations are specific to myofibroma.6 A subsequent analysis by Hung et al. revealed that 80% of myopericytomatoses, a diffuse and multinodular variant of myopericytoma, and 60% of conventional myopericytomas harbored PDGFRB mutations.7 Although the reason for inconsistency in determining the PDGFRB mutation status in pericytic tumors remains unclear, it may be attributed to the differences in diagnostic criteria of pericytic tumors with potential combined morphology. NOTCH3 mutations have also been reported in familial infantile myofibromatosis,4 and to a lesser extent in sporadic myofibroma,8 however, mutation status in other types of pericytic tumor remains unclear. Therefore, we examined 41 pericytic tumors, including combined tumors, by PCR using FFPE tumor tissues and primers for detecting hot spot mutations in exon 12 and 14 of PDGFRB 6,7 and exon 25 of NOTCH3.4,8 Our results show these mutations to be present in myopericytomas (PDGFRB: 27%, NOTCH3: 36%), myofibromas (50%, 0%), angioleiomyomas (15%, 23%), and glomus tumors (56%, 11%). Four PDGFRB mutation-positive lesions, myopericytoma-angioleiomyoma, myopericytoma-myofibroma, and myofibroma-/angioleiomyoma-myopericytoma, were identified as combined pericytic tumors. Three NOTCH3 mutation-positive lesions, myopericytoma-angioleiomyoma, myopericytomatosis-myofibroma, and angioleiomyoma-myopericytoma, were also confirmed as combined pericytic tumors. We identified PDGFRB and NOTCH3 mutations in a wider variety of pericytic tumors than that described in previous studies, including angioleiomyoma, glomus tumor, and combined tumor. Our present study describes PDGFRB mutations as having relatively greater frequency in myofibroma and glomus tumors, and NOTCH3 mutations in myopericytomas. However, the overall results indicate that detection of PDGFRB or NOTCH3 mutations is neither specific nor useful for subclassification of tumors belonging to the pericytic tumor family. PDGFRB mutations were also detectable in angioleiomyoma and glomus tumors, but were less frequent in myopericytoma and myofibroma than those detected in previous studies.6,7 These discrepancies in results could be partly attributed to our research design, wherein the primers used cover the hot spots in exon 12 and exon 14 of PDGFRB, but not other parts, such as exon 11 or exon 18, which have been reported to harbor certain mutations.8,28
With regard to biological functions, PDGFRB has a crucial role in the development of vascular smooth muscle cells and pericytes. This role has been demonstrated by analyses of mutations such as R561C in exon 12, which is considered to compromise the autoinhibitory role of the juxtamembrane domain and consequently lead to downstream signaling-activation.3 Other mutations, such as P660T and N666K, are recurrently observed in myofibromas or myopericytomas,3–7 some of which have been considered activating mutations in biochemical studies.28 NOTCH signaling also plays an important role in vascular development or homeostasis, as well as PDGF signaling.29 In vascular smooth muscle cells, NOTCH and PDGF share a common signaling pathway to control their vascular differentiation by an NICD binding to alternative CSL sites of PDGFRB promoter in response to ligand stimulation.30 The NOTCH3 L1519P mutation, located in exon 25, has been reported in infantile myofibromatosis,4 and to a lesser extent in myofibroma, along with low-level copy number gains.8 Exon 25 of NOTCH3 encodes the negative regulatory region, encompassing LIN12-Notch repeat and heterodimerization domain, which are considered to maintain receptor quiescence by preventing protease cleavage prior to ligand binding.31 Previous studies demonstrated that NOTCH3 L1519P mutations, located in the NOTCH3 negative regulatory region, enhance ligand-independent signaling and PDGFRB expression.32,33 Thus, NOTCH3 mutations could be potentially be associated with pericytic tumor development by means of a mechanism similar to that of PDGFRB mutations.
Our immunohistochemical results showed variable expression levels of PDGFRB and NOTCH3 in pericytic tumors. NOTCH3-mutated tumors were frequently positive for nuclear NOTCH3, which indicates activated NOTCH3 signaling. This was significantly different to nuclear NOTCH3 levels in tumors without NOTCH3 mutations, suggesting mutation-induced aberrant NOTCH3 expression. Immunohistochemical analysis of PDGFRB-mutation status revealed a significant difference in immunohistochemical positivity between the PDGFRB mutant and non-mutant groups. Four NOTCH3-mutated tumors were also positive for PDGFRB despite lacking PDGFRB mutation, which might indicate that NOTCH signaling upregulates PDGFRB expression in vascular smooth muscle cells.30,33 Therefore, PDGFRB immunohistochemical positivity may not always correspond with PDGFRB gene alteration. Some tumors with neither PDGFRB nor NOTCH3 mutations were positive for PDGFRB and/or nuclear NOTCH3 immunohistochemically. Such discordant genetic and immunohistochemical results associated with pericytic tumors should be addressed by additional studies.
Recent studies have demonstrated SRF or NOTCH gene rearrangements to be distinct molecular genetic abnormalities associated with pericytic tumors.9–13 SRF gene rearrangements, such as gene fusions with RELA, CITED1, CITED2, NFKBIE, and NCOA2, have been identified in perivascular myoid tumors, including cellular variants of myofibroma/myopericytoma or their combined tumors.9,10 Previous studies have identified NOTCH1-3 gene rearrangements in 54% of glomus tumors, 0–18% of myopericytomas, and 6–11% of angioleiomyomas, but not in myofibromas.11,12 NOTCH2 gene rearrangements appeared to be the most common abnormality, accounting for 88% of NOTCH1-3 gene rearrangements. MIR143 is reported to be a fusion gene partner of NOTCH, and its strong promoter leads to oncogenic overexpression of NOTCH2.11 CARMN is also described as fusion partner of NOTCH2 in glomus tumors of the upper digestive tract.13 Our FISH analysis did not reveal SRF rearrangement in the pericytic tumors examined, but identified MIR143/NOTCH2 gene fusion in two glomus tumors. None of the examined tumors were found to exhibit simultaneous gene mutations and rearrangements, supporting the notion that PDGFRB/NOTCH3 mutations and SRF or NOTCH rearrangements are mutually exclusive molecular genetic abnormalities. However, the frequency of NOTCH2 rearrangements of glomus tumors in our study was lower than previously reported. This discrepancy may be due to the research design, such as smaller sample size and variation in race of subjects. Alternatively, although MIR143 is currently considered the most common fusion partner of NOTCH2 in glomus tumor of soft tissue, this may be due to the fact that we did not search for partners other than MIR143 with MIR143/NOTCH2 translocational FISH probes.
Our study has several limitations. Although we observed a diverse range of PDGFRB and NOTCH3 mutations, they mostly included seldom described minor mutations such as PDGFRB S574F, M655I, and H657L mutations and NOTCH3 A1480S/T, D1481N, G1482S, E1491K, T1490A, and G1494S mutations, some of which are listed in the NCBI dbSNP database as pathologically uncertain variants at very low frequencies. Previous studies have shown a frequent occurrence of minor mutations such as PDGFRB W566S, I569V, Y589N, and K653E in somatic pericytic tumors,6–8, 28 but the pathogenic role of these remains uncertain. However, the minor mutations detected in our study, such as PDGFRB S574F, M655I, and H657L, and NOTCH3 G1482S, T1490A, and G1494S, were classified as ‘deleterious/damaging’ by more than four pathogenicity prediction tools in silico, suggesting a possible pathogenetic role in tumorigenesis of pericytic tumors. Multiple mutations were found in some pericytic tumors, such as a glomus tumor having PDGFRB G576S, M655I, and N666K, and an angioleiomyoma having PDGFRB N666K and NOTCH3 D1481N and G1494S. Some of the somatic myofibromatoses previously described also harbored multiple mutations such as G576D/N666K and R561C/N666K.6,8 In particular, R561C and N666K, both considered activating mutations, were recurrently detected in myofibroma or myopericytoma. A recent study described that 30% of lung adenocarcinomas harbor two or more driver mutations, such as EGFR, TP53, SETD2, or SMARCA4.34 Based on our present study findings, PDGFRB and NOTCH3 mutations might be considered to work cooperatively in pericytic or vascular smooth muscle differentiation and/or tumorigenesis, but our sample size is too limited to conclusively determine the true function of gene mutations as drivers or passengers using pathogenicity prediction tools only. Further molecular and biological studies are necessary to elucidate the pathogenetic role of such minor mutations.
In conclusion, we identified PDGFRB and NOTCH3 mutations in a wider variety of pericytic tumors, including combined tumors, than that reported in previous studies. We showed that myopericytomas, myofibromas, angioleiomyomas, and glomus tumors are among the members of the pericytic tumor family that harbor PDGFRB and/or NOTCH3 mutations. These results indicate that detection of PDGFRB or NOTCH3 mutations is not particularly useful for subclassifying pericytic tumors. Immunohistochemistry reveals NOTCH3 mutations are associated with positive staining for nuclear NOTCH3 and PDGFRB expression, which are both potentially involved in the PDGF signaling pathway. Consequently, PDGFRB immunoreactivity cannot always be indicative of PDGFRB mutations in pericytic tumors.