Mosaic RASopathies

Abstract

“RASopathies” are a group of developmental syndromes with partly overlapping clinical symptoms that are caused by germline mutations of genes within the Ras/MAPK signaling pathway. Mutations affecting this pathway can also occur in a mosaic state, resulting in congenital syndromes often distinct from those generated by the corresponding germline mutations. For syndromes caused by mosaic mutations of the Ras/MAPK signaling pathway, the term “mosaic RASopathies” has been proposed. In the following article, genetic and phenotypic aspects of mosaic RASopathies will be discussed.

The RAS Signaling Pathway

Ras proteins are small GTPases playing a key role in transducing extracellular growth factor stimuli into the intracellular environment. After the binding of growth factors to their corresponding receptors on the cell surface, guanine-nucleotide exchange factors (e.g., SOS proteins) are activated. These factors cause a switch from inactive GDP-Ras to active GTP-Ras. On the other hand, GTPase-activating proteins, such as neurofibromin and p120GAP, catalyze the hydrolysis of GTP-Ras to inactive GDP-Ras, thus terminating Ras signaling.¹ Active GTP-Ras can cross-talk with further signaling proteins and stimulate downstream pathways. The two most important Ras-dependent pathways are the Ras-Raf-MEK-ERK and the PI3K-Akt signaling pathways. Activation of these pathways finally influences cell survival/apoptosis, proliferation and differentiation in various tissues. Three genes encode for the classical Ras proteins: HRAS, KRAS and NRAS. The protein sequences of the different Ras proteins have been highly conserved in phylogenesis, especially the N-terminal G domain, whereas the C-terminal sequence that is important for the targeting of Ras proteins to the cell membrane show a higher degree of variability.²

Considering the central position of Ras proteins in cellular signaling processes, it is not surprising that the disturbance of Ras protein function is fundamentally involved in the pathogenesis of many human disorders; e.g., activating mutations in RAS genes are found in about 30% of human cancers.³ These oncogenic mutations are somatic alterations, thus restricted to the tumor tissue and absent in non-neoplastic tissues of the patients. According to current concepts, they are most likely acquired during life, e.g., by exposure to mutagenic factors. Interestingly, different cancer types often show the specific involvement of one of the three RAS genes, and the activating mutations occur preferentially at hotspot loci of codons 12, 13 and 61.

Germline RASopathies

In addition to the somatic mutations in cancer, mutations of RAS genes or other components of the Ras-Raf-MEK-ERK signaling pathway can also occur in germline. These germline mutations result in developmental syndromes, with craniofacial dysmorphology and abnormalities of the heart, skin, eyes, brain and musculoskeletal system. Furthermore, some of these patients harbor an increased risk for the development of cancer during life. The term “RASopathies” has been suggested for congenital syndromes caused by germline mutations in genes of the Ras-Raf-MEK-ERK signaling pathway.⁴ This class of genetic syndromes shows a considerable phenotypic overlap, which is explained by the common pathologic activation of the same pathway. Some RASopathies even show mutations of the same gene. The RASopathies currently comprise Noonan syndrome (PTPN11, SOS1, KRAS, NRAS, RAF1), neurofibromatosis type 1 (NF1), LEOPARD syndrome (PTPN11, RAF1), hereditary gingival fibromatosis (SOS1), capillary malformation-arteriovenous malformation syndrome (RASA1), Costello syndrome (HRAS), autoimmune lymphoproliferative syndrome (NRAS), cardio-facio-cutaneous syndrome (BRAF, KRAS, MAP2K1, MAP2K2) and Legius syndrome (SPRED1). In most cases, the mutations in RASopathies are heterozygous, result in an activation of the Ras/MAPK signaling pathway and follow an autosomal dominant inheritance. Though there may be some overlap (e.g., HRAS p.G12S is found both in Costello syndrome and in sporadic cancer), in most cases the mutational spectrum of the affected genes is different between germline RASopathies and sporadic cancer. While some RASopathies are rather frequent, such as the Noonan syndrome (1/1,000 to 1/2,500 newborns) or neurofibromatosis type 1 (1/3,000 newborns), others are rare disorders such as Costello and LEOPARD syndrome. However, some RASopathies are caused by inactivating mutations. RASA1 acts like a tumor-suppressor gene,^5,6 promoting the conversion of active GTP-bound Ras to inactive GDP-bound Ras. It has been hypothesized that in the background of a germline RASA1 mutation, a second hit in the wild-type allele randomly occurring at different body sites causes the vascular anomalies, according to the two-hit hypothesis of Knudson.

Mosaic RASopathies

Genetic mosaicism is defined by the presence of at least two genetically distinct cell populations in the same organism. It results from postzygotic mutations. The clinical phenotype of mosaic disorders is determined by the timing of mutation, the level of pathway activation as well as the affected cell type. Happle has hypothesized that the phenotypical consequences of some activating mutations in the germline may be so fundamental that they are not compatible with development and life. Therefore, these mutations may only survive in a mosaic state with a limited number of affected tissues and cells.⁷ For some of the established RASopathies, like Costello syndrome or neurofibromatosis type I, patients with mosaic mutations have been described. In these cases, the phenotype was not significantly different from the germline variants. Apart from that, mosaic mutations affecting the Ras-Raf-MEK-ERK signaling pathway have been identified in congenital cutaneous disorders like epidermal nevi. The fact that they show a distinct phenotype compared with the germline RASopathies prompted us to propose the designation “mosaic RASopathies” for this group of congenital disorders (Table 1).⁸

Table 1. Mosaic RASopathies.

Mosaic RASopathy	Gene	Reported mosaic mutations	References
Keratinocytic epidermal nevus	HRAS KRAS NRAS (FGFR3)	G12C, G12V, G13R, Q61L G12D G12D, P34L, Q61R R248C, S249C, G372C	8 8 8 8 , 14 , 45
Keratinocytic epidermal nevus syndrome	HRAS KRAS (FGFR3)	G12S G12D R248C, S249C	19 18 20 - 23
Sebaceous nevus	HRAS KRAS	A11S, G12C, G12D, G12S, G13R G12D, G12V	25 25
Schimmelpenning syndrome	HRAS KRAS	G13R G12D	25 25
Mosaic Costello syndrome	HRAS	G12S	32 , 33
Segmental neurofibromatosis type I	NF1	Various point mutations, deletions, insertions	38 , 39

Keratinocytic epidermal nevi

Epidermal nevi represent a paradigm for cutaneous mosaic disorders.^9,10 The lesions are congenital and are arranged in a linear fashion following Blaschko’s lines. They may be either visible at birth or become manifest during the first years of life. According to the affected skin components, epidermal nevi have been divided into organoid (with abnormal adnexal components, such as the hair follicles, sebaceous and sweat glands) and non-organoid (keratinocytic) types (with only epidermal changes).¹¹ The common keratinocytic epidermal nevus is the most frequent non-organoid epidermal nevus type.¹² It is benign, although in some cases, development of malignant tumors such as basal cell carcinomas and squamous cell carcinomas on the basis of a pre-existing keratinocytic epidermal nevus has been reported. A part of these nevi is caused by mosaic FGFR3 and PIK3CA mutations.^13,14 More recently, it has been shown that common keratinocytic epidermal nevi also result from mosaic RAS mutations.⁸ In a series of 72 nevi, RAS mutations were found in 39% of the lesions. The most frequent mutations occurred in the HRAS gene, with the G13R mutation representing a hotspot mutation. In addition, KRAS and NRAS mutations were identified. In some samples, the HRAS mutation co-occurred with a PIK3CA mutation, while RAS and FGFR3 mutations were mutually exclusive. The RAS mutations were present in a mosaic state, as skin tissue adjacent to the epidermal nevus and blood leukocytes of the patients showed a wild-type sequence. These findings prove a close genotype-phenotype association and incorporate common keratinocytic epidermal nevi to the mosaic RASopathies. Because the identified FGFR3 point mutations in keratinocytic epidermal nevi have been shown to activate, at least in part, the Ras-Raf-MEK-ERK signaling pathway in keratinocytes and urothelial cells,^15,16 keratinocytic epidermal nevi caused by FGFR3 mosaic mutations may also be added to the group of mosaic RASopathies. This classification is further supported by the fact that FGFR3 mutant and RAS mutant keratinocytic epidermal nevi show an undistinguishable clinical and histologic phenotype.

Keratinocytic epidermal nevus syndrome

Epidermal nevus syndromes are defined by the association of an epidermal nevus with extracutaneous abnormalities, such as cerebral, musculosceletal, cardiac, renal and ocular defects.¹⁷ Of note, this definition includes cases in which extracutaneous tissues are affected by the same mosaic mutation as the epidermal nevus, however, do not show an aberrant phenotype. The keratinocytic epidermal nevus syndrome in particular is characterized by a keratinocytic epidermal nevus, whose underlying mosaic mutation has spread to other organs. Although the exact incidence of this syndrome is unknown, it is less frequent than keratinocytic epidermal nevi without involvement of other organs. The corresponding extracutaneous abnormalities comprise neuronal defects, such as seizures, mental retardation, hemimegalencephaly, ventricular abnormalities, cortical atrophy and hemiparesis. Skeletal manifestations include incomplete bone formation, hypertrophy or hypoplasia of bones, bone cysts, kyphoscoliosis and vitamin D-resistant rickets.

Recently, RAS mutations have been identified in patients diagnosed with keratinocytic epidermal nevus syndrome. For instance, the co-occurrence of a keratinocytic epidermal nevus, a rhabdomyosarcoma, polycystic kidneys and growth retardation in an infant could be attributed to an oncogenic mosaic KRAS G12D mutation.¹⁸ In a second patient with a systematized keratinocytic epidermal nevus, who had developed multiple urothelial carcinomas, a mosaic HRAS G12S mutation was identified in the keratinocytic epidermal nevus, the urothelium, the urothelial cell carcinomas and in the blood leukocytes.¹⁹ Both patient cases are examples of mosaic RASopathies. Further studies will have to determine the prevalence of mosaic RAS mutations in keratinocytic epidermal nevus syndrome. Besides RAS mutations, the keratinocytic epidermal nevus syndrome can also be caused by FGFR3 mosaic mutations.^20-23 The reported patients demonstrate mild facial dysmorphism, cerebral abnormalities with seizures and mental retardation, scoliosis and involvement of blood leukocytes. In analogy to the FGFR3-mutant keratinocytic epidermal nevi, the corresponding FGFR3-mutant keratinocytic epidermal nevus syndromes can be designated as mosaic RASopathies. It will be interesting to analyze in a larger cohort of patients whether the different RAS and FGFR3 mutations in keratinocytic epidermal nevus syndrome show a distinct clinical spectrum of symptoms. Both patients with a RAS-mutant keratinocytic epidermal nevus syndrome developed cancers (rhabdomyosarcoma and urothelial carcinoma), whereas the four reported patients with a FGFR3-mutant keratinocytic epidermal nevus syndrome had no history of cancer. It will be particularly important for the prognosis of the patients and clinical surveillance programs whether this association is significant or occurred by chance.

Sebaceous nevus

Sebaceous nevi are organoid epidermal nevi that are preferentially localized in the head and neck region. Histologically they are characterized by abundant sebaceous glands, epidermal hyperplasia and apocrine elements.²⁴ A peculiar feature of sebaceous nevi is the secondary development of mostly benign tumors in about 25% of lesions during life. These nevi present clinically as hairless, yellow-orange plaques of varying size and shape. Recently it has been shown that sebaceous nevi are caused by mosaic mutations of HRAS and KRAS genes.²⁵ HRAS mutations were found in 95%, and KRAS mutations in 5%, of the lesions. Some nevi showed double mutations of RAS genes. In total, 97% of sebaceous nevi harbored a RAS mutation. Analysis of mutiple non-lesional tissues revealed a wild-type sequence, indicating mosaicism. The HRAS G13R mutation was the predominant mutation present in 91% of the nevi. Cell culture of lesional keratinocytes demonstrated that these cells are the carrier of the HRAS mutation. Since sebaceous nevi are rather frequent with an incidence of approximately 1 in 1,000 live births, and RAS mutations have been found in a high percentage of the lesions, sebaceous nevi probably represent the mosaic RASopathy with the highest prevalence.

Schimmelpenning syndrome

In analogy to the keratinocytic epidermal nevi, sebaceous nevi can be associated with extracutaneous abnormalities, such as cerebral, ocular and skeletal defects. This syndrome is called Schimmelpenning syndrome or linear sebaceous nevus syndrome.²⁶ In a large series of 196 patients with sebaceous nevus, 7% revealed neurological abnormalities like mental retardation, seizures and hemimegalencephaly.²⁷ Ocular abnormalities comprise coloboma and lipodermoids, and skeletal defects include hypoplastic bones, short stature, incomplete formation of bony structures and vitamin D-resistant hypophosphatemic rickets, similar to the keratinocytic epidermal nevus syndrome. Two patients with Schimmelpenning syndrome^28,29 were recently analyzed, and both exhibited RAS mosaic mutations.²⁵ One patient was a 52-y-old woman revealing a sebaceous nevus on her scalp, neck, trunk and right arm, which was associated with growth retardation, ocular abnormalities, disproportionate hyposomia, multiple bone fractures and bone deformation due to hypophosphatemic rickets. In this patient, analysis of multiple lesional tissues showed the HRAS G13R mutation, which was absent in blood leukocytes, suggesting a mosaicism of the HRAS mutation. The second subject with Schimmelpenning syndrome was a monozygotic twin. While his brother showed no abnormalities, the patient displayed a severe Schimmelpenning syndrome, with a sebaceous nevus on the face, ocular abnormalities, an isolated cleft palate, a patent ductus arteriosus and cerebral defects. Analysis of a biopsy from the sebaceous nevus of this subject showed the KRAS G12D mutation, which was absent in normal skin and blood leukocytes, thus confirming mosaicism. Therefore, Schimmelpenning syndrome can be considered as a mosaic RASopathy, although the frequency of mosaic RAS mutations in this syndrome has to be further evaluated in a larger cohort of patients.

Costello syndrome

Costello syndrome is a very rare RASopathy caused by HRAS germline mutations.³⁰ It is characterized by growth retardation, skin changes, including deep palmar and plantar creases, papillomata, loose skin, abnormal fingernails, spatulate finger pads and increased pigmentation, cardiomyopathy, coarse face and cancer predisposition.³¹ Two patients with a mosaic Costello syndrome have been reported in the literature, both of them being caused by a mosaic HRAS mutation. One patient displayed a phenotype suggestive for Costello syndrome, however, lacked an HRAS mutation in her blood leukocytes.³² In contrast, analysis of buccal swabs showed that approximately 30% of the cells carried an HRAS G12S mutation, thus confirming mosaicism of this mutation. Another report described a father who had a mosaic HRAS G12S mutation.³³ This mutational mosaicism obviously involved the gonads of the subject, because he had an offspring with an HRAS G12S germline mutation and Costello syndrome, suggesting a father-to-son transmission of the mutation. Both reports provide evidence that in rare cases, Costello syndrome may manifest as a mosaic RASopathy.

Segmental neurofibromatosis type I

Neurofibromatosis type I is an autosomal dominant inherited RASopathy with an incidence of 1 in 2,500 to 1 in 3,000 individuals.³⁴ The clinical findings comprise cafe-au-lait spots, neurofibromas of the skin, Lisch nodules in the eye and plexiform neurofibromas. Furthermore, this syndrome is associated with abnormalities of the vascular, skeletal and central nervous system. Of note, patients with neurofibromatosis type I have an increased risk to develop benign and malignant tumors, such as optic pathway gliomas or malignant peripheral nerve sheath tumors. Neurofibromatosis type I is caused by germline mutations of the NF1 gene, which encodes for the tumor-suppressor protein neurofibromin. In contrast to many other RASopathies that are characterized by activating heterozygous mutations, the mutations in the NF1 gene are inactivating. According to the two-hit hypothesis of Knudson, the second wild-type allele of NF1 is inactivated by somatic mutations, small deletions or insertions (loss of heterozygosity).^35,36 Some Schwann cells in neurofibromas of neurofibromatosis type I patients show loss of heterozygosity of NF1, while the other cells still retain one functional wild-type allele.³⁷ According to these findings, benign and malignant tumors, cafe-au-lait spots and other lesions in neurofibromatosis type I are presumably the result of inactivation of both NF1 alleles, one by an inherited mutation and the other by a second genetic alteration. The variability of the phenotype of the lesions and their localization in neurofibromatosis patients might be explained by the element of chance in determining what cell types are involved by the second hit and at which localization this happens. Interestingly, some abnormalities observed in neurofibromatosis type I, such as general hyperpigmentation of the skin, are obviously caused by the loss of one NF1 allele (haploinsufficiency).

Somatic mosaicism of NF1 deletions has been reported in two patients with generalized neurofibromatosis type I that were phenotypically indistinguishable from patients with germline mutations.^38,39 Moreover, some patients display a segmental form of neurofibromatosis type I. These patients may have pigmentary changes only, neurofibromas only, both pigmentary changes and neurofibromas or isolated plexiform neurofibromas that are restricted to a segmental area of the body.⁴⁰ Patients with segmental neurofibromas interestingly show a neural distribution in dermatomes, because the genetic mutation appears to be restricted to Schwann cells. In contrast, patients with solely pigmentary segmental changes show a distribution that follows the lines of Blaschko, and melanocytes have been identified as an affected cell type. Segmental neurofibromatosis type I is thought to be about 30 times less frequent than the germline variant.⁴¹ The segmental neurofibromatosis type I results from postzygotic genetic alterations in the NF1 gene such as microdeletions.⁴² A second hit in a cell within the affected segment will cause a lesion (e.g., neurofibroma), whereas inactivation of an NF1 allele outside this segment will not result in a lesion because of the remaining wild-type allele. All cases of postzygotic NF1 alterations (“first hit“) that lead to mosaicism can be categorized as mosaic RASopathies. The mosaicism might result in a segmental neurofibromatosis type I, but also in a generalized disease indistinguishable from germline neurofibromatosis, depending on the time point at which the mutation occurs during embryogenesis.

Other Mosaic RASopathies

Further established RASopathies may occur in mosaicism in rare cases. For example, a patient with LEOPARD syndrome has been published whose left trunk and arm were devoid of the classical lentigines observed on other body sites.⁴³ This phenotype suggests a mosaicism, but the underlying mechanism was unknown, as both fibroblasts from lesional and non-lesional skin areas harbored the PTPN11 mutation. Silencing of the mutated gene by an unidentified second mutation or by epigenetic factors as well as revertant mosaicism may account for the observed phenotype in this case.

Functional Aspects

Most of the mutations underlying mosaic RASopathies are point mutations resulting in an activation of the Ras-Raf-MEK-ERK signaling pathway. These mutations can be designated “oncogenic,” as they are also found in a variety of benign and malignant tumors and have been shown to be tumorigenic in vitro and in vivo.⁴⁴ Interestingly, the mutation spectrum of mosaic RASopathies shows a considerable overlap with that of tumors but is often different from germline RASopathies. Another interesting fact is that in several mosaic RASopathies, the mutational spectrum is characterized by one dominant hotspot mutation. For example, common keratinocytic epidermal nevi show a predominance of the HRAS G13R and the FGFR3 R248C mosaic mutations.^8,14,45,46 Furthermore, the HRAS G13R mosaic mutation is found in approximately 90% of sebaceous nevi.²⁵ The reason for this predominance of one single hotspot mutation remains unknown. Carcinogens with a site-specific mutagenic effect in embryogenesis may explain this phenomenon. Another possibility is that in comparison to other mutations occurring during embryogenesis, hotspot mutations provide a growth advantage for the affected cell clones, thus resulting in a positive selection of the respective clones. In contrast to the mosaic RASopathies, the spectrum of somatic mutations of the same genes in tumors is more heterogeneous. For example, in seborrheic keratosis, a benign epidermal skin tumor that is histologically almost identical to keratinocytic epidermal nevus, the spectrum of FGFR3 and RAS mutations is considerably more diverse than in the epidermal nevi.^47-49

For some genes such as HRAS, the observed phenotypical pleiotropy of the same mutation in mosaic RASopthies vs. germline RASopathies is intriguing. The HRAS G12S mutation, for example, is the most frequent germline mutation in Costello syndrome.³¹ The same mutation has been identified in sebaceous nevi²⁵ and in a case of keratinocytic epidermal nevus syndrome.¹⁹ Though patients with Costello syndrome show a skin phenotype,⁵⁰ it is markedly different from that observed in mosaic RASopathies. We have hypothesized that the cell type affected by the mutation (e.g., both epithelial and mesenchymal cells in Costello syndrome vs. epithelial cells in epidermal nevi), as well as the time point at which the mutation occurs in embryogenesis, may be critical determinants for the observed phenotypic difference between germline and mosaic RASopathies.

Moreover, even in mosaic RASopathies, the observed pleiotropy is remarkable. The mosaic HRAS G13R mutation, for example, can result both in a sebaceous nevus and a keratinocytic epidermal nevus, the latter lacking abundant sebaceous glands and the dilated apocrine glands that are observed in sebaceous nevi.^8,25 We hypothesize that this variability of phenotypes caused by an identical mosaic mutation is best explained by the varying differentiation potential of a mutated progenitor cell. The differentiation of cells is a complex process that is strictly organized and controlled depending on the developmental stage of the organism and the body site. Keratinocytic progenitor cells in the head and neck region, for example, will differentiate into epidermal keratinocytes and sebaceous glands, whereas other body sites show less sebaceous glands. Thus, if the mosaic HRAS mutation affects a progenitor cell in the head and neck region with a sebaceous differentiation potential, a sebaceous nevus might result. If the mosaic mutation occurs in a progenitor cell of the trunk, the sebaceous differentiation program will not be activated, because skin at this region contains less sebaceous glands, and the resulting phenotype will be a keratinocytic epidermal nevus without abundant adnexal structures. According to this concept, the cellular context of the mosaic mutation (i.e., regulatory mechanisms that depend on the cell type, the stage of embryonic development and the specific anatomical region) will determine the resulting phenotype of mosaic RASopathies.

Mosaic RASopathies and Cancer

The Ras-Raf-MEK-ERK signaling pathway has a central role in tumorigenesis, with approximately 30% of human tumors harboring oncogenic RAS mutations.^3,51 Other genes of this pathway are also reported to show genetic alterations in benign and malignant tumors such as BRAF in malignant melanoma⁵² and in melanocytic nevi.⁵³ Therefore, it is not surprising that several RASopathies are associated with an increased risk of developing (malignant) tumors during life.⁵⁴ This association between RASopathies and cancers is biologically plausible, as both germline and sporadic RAS mutations activate the same pathway. In a retrospective analysis, 4% of patients with Noonan syndrome were found to develop various cancers, such as myeloproliferative disease, neuroblastoma, low-grade glioma, rhabdomyosarcoma and acute lymphoblastic leukemia. In Costello syndrome, 10% of patients had a history of cancer, including rhabdomyosarcoma, bladder cancer and neuroblastoma. A cancer incidence peak in childhood was observed in both syndromes.⁵⁴ In the same study, 4% of patients with cardio-facio-cutaneous syndrome had been identified with acute lymphoblastic leukemia, non-Hodgkin lymphoma, rhabdomyosarcoma and hepatoblastoma. In neurofibromatosis type I, the frequent development of neurofibromas is a criterion for the diagnosis of the syndrome. The risk of malignant myeloid disorders in children with neurofibromatosis type I is 200‒500 times the normal risk.⁵⁵ Moreover, these patients have a higher susceptibility for malignant peripheral nerve sheath tumors,⁵⁶ optic pathway gliomas,⁵⁷ pheochromocytomas⁵⁸ and other malignancies.

Mosaic RASopathies also harbor an increased risk for tumor development. Of the two reported patients with an epidermal nevus syndrome caused by a RAS mutation, one patient developed a rhabdomyosarcoma in childhood.¹⁸ A postzygotic KRAS G12D mutation was identified as the underlying mutation, suggesting that mosaicism of this genetic alteration caused the keratinocytic epidermal nevus and contributed to the rhabdomyosarcoma. This assumption is compatible with the reported dysregulation of the RAS signaling pathway in human rhabdomyosarcoma.⁵⁹ The other patient developed bladder cancer at the unusually young age of 19 y.¹⁹ After a period of 29 y, the patient was again found to have urothelial cancer of the bladder and the renal pelvis. Furthermore, a metastasis of the urothelial carcinoma was found in the lung. The HRAS G12S mutation was detected in the epidermal nevus tissue, the urothelial carcinomas, the lung metastasis as well as in the normal urothelium, whereas it was absent in the muscle layer of the bladder, thus confirming mosaicism. These examples demonstrate that congenital mosaic mutations of oncogenes may predispose to cancer in children and adults. Malignant tumors that occur in young patients without a history of familial cancer or tumors that develop in a multicentric manner may thus indicate congenital mosaicism of an oncogenic mutation.

An association between epidermal nevus syndromes and cancers, such as urothelial carcinoma and rhabdomyosarcoma, had already been known from the literature.^60-64 However, it remains unknown whether these cases were caused by mutations of the RAS signaling pathway, as these reported syndromes have not been genetically analyzed. Further studies are necessary to determine the incidence of malignancies in epidermal nevus syndromes and a possible correlation between the underlying gene mutations and specific cancer types. The risk for the development of a malignant tumor might depend on the activation potential of the respective mutation, the size of the mutated mosaic patch and the tissue and cell types affected by the mutation. We hypothesize that even though the mosaicism can be rather widespread and involve many different organs, only specific tissue types may be prone to develop cancer in the presence of a predisposing mosaic mutation.

Approximately 25% of sebaceous nevi develop secondary tumors during life.²⁴ The majority of these tumors is benign, comprising mainly trichoblastoma, syringocystadenoma papilliferum and further benign adnexal tumors, but malignant tumors have also been reported.^65-67 These tumors derive directly from sebaceous nevus cells, as they were shown to carry the same HRAS mutation as the underlying nevus.²⁵ Therefore, RAS mutations in sebaceous nevi may predispose to the development of benign and malignant tumors. Although the exact mechanisms remain elusive, either a second genetic hit or other pathogenetic factors could foster the tumor growth in the context of constitutive RAS pathway activation by the mosaic mutation. This view is supported by the fact that RAS mutations are associated with keratoacanthomas and squamous cell carcinomas that develop in patients with metastasized malignant melanomas receiving a therapy with a BRAF inhibitor. While the prevalence of RAS mutations in sporadic human squamous cell carcinomas is rather low,⁶⁸ RAS mutations are significantly more frequent in these tumors developing under BRAF inhibitor therapy.⁶⁹ These findings suggest a model of tumor growth by paradoxical pathway activation by BRAF inhibitors in RAS-primed mutant keratinocytes. This hypothesis is supported by a mouse model in which tumor growth of HRAS mutant keratinocytes was not initiated but accelerated by a BRAF inhibitor.⁷⁰ There are also a few reports of neoplasms growing on a pre-existent keratinocytic epidermal nevus.⁷¹ However, it is unknown why the prevalence of secondary tumors in sebaceous nevi is considerably higher than in keratinocytic epidermal nevi. Once more, the mutant cell type (epidermal keratinocytic stem cell vs. sebaceous stem cell), paracrine factors as well as varying exogenous factors, resulting in additional genetic hits, may play a role for the observed difference.

Concluding Remarks

Mosaicism is an important contributor to human disease, and mosaic RASopathies have emerged as a new class of congenital disorders. Though mosaic RASopathies share a common pathogenesis and show some clinical overlap with germline RASopathies, the occurrence of the mutations in a mosaic state often leads to a phenotype that is not merely an incomplete manifestation of the corresponding germline RASopathy, but results in a distinct clinical entity such as epidermal nevi. For some genes, like HRAS, the associated mosaic RASopathies (keratinocytic epidermal nevi and sebaceous nevi) are considerably more frequent than the corresponding germline RASopathy (Costello syndrome). The knowledge of the clinical and pathogenetic characteristics of mosaic RASopathies is important for physicians. In contrast to classical germline genetic disorders, the diagnosis often cannot be made by analysis of blood DNA, because in many patients, the mosaicism does not involve the bone marrow. It is crucial to analyze lesional tissue from the patients for the detection of the underlying mosaic mutation, which sometimes may be challenging or even impossible. While mosaic disorders of the skin can be analyzed rather easily by a biopsy due to the good visibility and accessibility of the lesions, this may be not the case for abnormalities of other organs, such as the brain or the skeletal system. The identification of the mutations underlying mosaic RASopathies help to classify the disorders and to make the exact diagnosis in patients with equivocal clinical symptoms. It might also be helpful for genetic counseling of the patients and their relatives. However, estimation of the risk for transmission of the mosaic mutation to the next generation is very difficult, because an involvement of the gonads by the mosaicism usually cannot be excluded. If the mosaic mutation is transmitted to the next generation, the offspring will harbor the mutation in all cells, which can be linked to a severe clinical phenotype that sometimes may be not compatible with life. For example, the mosaic FGFR3 R248C mutation that is frequently found in keratinocytic epidermal nevi is associated with thanatophoric dysplasia in germline,⁷² a severe skeletal dysplasia syndrome that is lethal at the time of birth or in the first years of life. The most frequent HRAS mutation in mosaic RASopathies, the G13R substitution, has not been described in germline up to now. It is unknown whether this gene mutation is lethal in germline. In contrast, other HRAS mutations, such as G12S, have been found in a mosaic state, for example, in sebaceous and keratinocytic epidermal nevi, but are also common in Costello syndrome patients. Thus, if an HRAS G12S mosaic mutation in a patient with epidermal nevus (syndrome) would affect the gonads and be transmitted to the offspring, the mutation would result in a classical Costello syndrome. The fact that most patients with epidermal nevi have healthy offspring suggests that in these patients, the risk of transmission is very low. However, the exact mechanisms of the distribution of the mutant cell clone during embryogenesis are not understood, and even a small visible mosaic skin lesion may not entirely rule out a gonadal involvement. In these cases, the identification of the underlying mutation in the mosaic (skin) lesion of the parents will allow a targeted prenatal genetic analysis.

Besides that, the knowledge of mosaic RASopathies is relevant due to their increased risk for the development of malignancies in children and adults. Depending on the affected tissues, as well as the respective gene mutations, these cancers may occur at an unusual age or multifocally. While the mosaic lesions in the skin are right before our eyes and can easily be monitored for the growth of tumors, this is not the case for internal organs such as the urinary tract. In patients with a widespread mosaic RASopathy, monitoring of organs that harbor a potential risk for cancer development might be recommended.

It is not unlikely that in the future, further mosaic RASopathies will be discovered. The predominance of mosaic RASopathies with a skin phenotype may be due to the better visibility of the lesions rather than to a true higher incidence. Mosaic RASopathies of internal organs may contribute to cancer and non-neoplastic disorders as well but remain unnoticed. The continuing improvement of genetic techniques (e.g., whole exome and deep sequencing approaches) will provide new insights in the prevalence and pathogenetic relevance of mosaic RASopathies in man in the near future.