Case report

We now report on an infant showing a large hyperkeratotic lesion present at birth and following a Blaschko's line, fully consistent with the diagnosis of EN (figure 1A). The child was born with normal weight, stature and head circumference (2.890 g, 49 cm and 33 cm, respectively) after an unremarkable pregnancy. She had no dysmorphy. The biopsy of the epidermal lesion confirmed the diagnosis by showing hyperkeratosis, acanthosis and papillomatosis (figure 1B). At 6 months of age, the girl developed an uterovaginal rhabdomyosarcoma with both embryonal and alveolar phenotypes (figure 1E). Surprisingly, the abdominal CT scan performed at the time of the tumour diagnosis showed bilateral micropolystic kidneys, a feature that was not found in the mother (no data from the father) (figure 1D). Moreover, the child failed to thrive (short stature −2DS, low weight <−2DS) linked to uncommon feeding difficulties that sustained after the end of chemotherapy (supplementary figure 1). No organic cause of this growth defect was found; in particular, the renal function was normal and no sign of intestinal malabsorption was detected. A skeleton x-ray examination was normal, and echocardiography showed no malformation. No developmental delay was noticed.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Phenotype and copy number analysis of the two tumours: (A) Depiction of the epidermal nevus following the Blaschko line. (B) Haematoxylin and eosin safran (HES) staining of the biopsy of the nevus: a, acanthosis; h, hyperkeratosis; p, papillomatosis. (C) Array-comparative genomic hybridisation. (CGH-array performed on Nimbelgen oligonucleotide array, using 50 ng DNA extracted from micro-dissected samples of the epidermal component of the biopsy specimen after whole genome amplification with phi29 DNA polymerase. Reference DNA was amplified according to the same procedures. No alteration is shown. (D) Abdominal CT scan showing the huge uterovaginal rhabdomyoasarcoma (→) and the micropolycystic kidney (▶). (E) HES staining of the tumour biopsy specimen in its alveolar subcomponent. (F) Array-CGH analysis of DNA from the tumour biopsy specimen: gain of whole chromosomes 2, 4, 8, 12, 13; amplifications (→) at 19p13.2, 19q13.2 and 22q11.21.

The unusual association of several clinical abnormalities prompted us to explore further the genetics of the EN and the rhabdomyosarcoma. Neither PIK3CA nor FGFR3 mutations were found in the DNA extracted from the nevus biopsy (Sanger method and SNaPshot methods, respectively). Considering that the tumour was of epithelial origin, we microdissected the hyperproliferative epithelium from the dermal tissue by laser capture-microdissection (Leica system, figure 2A,C). In order to search for a genetic lesion possibly causative for EN, DNA from the epidermal part was amplified by whole genomic amplification (phi29 DNA polymerase; New England Biolabs, Hitchin, UK); array-comparative genomic hybridisation (CGH) was then performed, using the oligonucleotide Nimbelgen array-CGH platform. As shown in figure 1C, no copy number alteration was found.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

G12D KRAS mutation in the epidermal component of the lesion. (A) Haematoxylin and eosin safran (HES) staining on the whole epidermal nevus biopsy; d, dermal component; e, epidermal part of the lesion. (B) Exon1 KRAS sequencing shows an unbalanced proportion of the normal allele and the G12D variant in DNA extracted from the whole section of the biopsy specimen. (C) Laser capture micro-dissection using a Leica microcapture system. The epidermal cells are isolated from the dermal tissue. After micro-dissection, the samples were pelleted in Qiagen buffer; DNA was extracted with QiaAmp DNA micro-kit (Qiagen, Courtaboeuf, France). 20 ng were used for direct PCR amplification and sequencing; 10 ng were amplified with phi29 DNA polymerase for subsequent array-comparative genomic hybridisation (CGH) (D) KRAS sequencing in the micro-dissected epidermal (e) and dermal (d) tissues. In the purely epidermal component, sequencing shows the heterozygous G12D mutation with a balanced peak for both normal and variant alleles. In contrast, no mutation is seen in the pure dermal component.

Despite the presence of an alveolar component, no FKHR translocation was found in the cDNA from the rhabdomyosarcoma. To more thoroughly characterise this atypical mixed tumour, genomic DNA was extracted from the primary biopsy and analysed by array-CGH. This analysis showed a gain of whole chromosome 8, an abnormality commonly seen in embryonal rhabdomyosarcomas, and other less common alterations (figure 1F).

Given the coexistence of four rare symptoms—that is, rhabdomyosarcoma, EN, micropolystic kidneys and failure to thrive with feeding difficulties, we postulated that the child may bare a single genetic lesion likely to account for the whole clinical phenotype. Knowing that (i) the mutations observed in epidermal nevi lead to a constitutive activation of the RAS-RAF cascade; (ii) mutations in the RAS pathway are occasionally detected in rhabdomyosarcomas and (iii) germline mutations in the RAS pathway predispose to both epidermal hyperproliferation and rhabdomyosarcomas, we explored this pathway by gene sequencing. Interestingly enough, whereas no mutation was found in NRAS or HRAS, an oncogenic G12D KRAS mutation was found in both the rhabdomyosarcoma and the epidermal component of the nevus (figure 2B,D). In contrast, no mutation was found in the dermal component of the skin lesion (figure 2D), or in a normal skin area; a wild-type sequence was also seen in blood lymphocytes, cheek swab and bone trephine biopsy specimens. No tissue was available to explore the status of KRAS in the polycystic kidney. To sensitise the detection of the G12D mutation, an allele-specific PCR was performed according to previously published methods.13 No variant was detected in the blood, bone biopsy and normal skin.

DISCUSSION

This report indicates for the first time that oncogenic KRAS mutation causes EN. Mutations in the RAS pathways are found in approximately 15% of rhabdomyosarcoma14–16; hence, the KRAS mutation in the malignant tumour could be more expected. However, most of the mutations involve the NRAS codon 61 and no mutation in KRAS has been found in the two most recent series of 31 and 30 primary rhabdomyosarcomas.14 15 Only one mutation in KRAS codon 12 has been reported.16 Although KRAS mutation has not been involved in polycystic kidney in humans, it is noteworthy that, in the mouse, selective hyperexpression of RAS in developing renal tubules results in the development of multiple renal cysts.17 This experimental observation provides strong support in favour of a causative role of the KRAS mutation in our patient's renal malformation.

Although we cannot formally rule out the possibility that the two identical KRAS alterations occur by chance, it has to be noted that KRAS mutations are rare in rhabdomyosarcoma, and that not a single G12D KRAS mutation in rhabdomyosarcoma is reported in the COSMIC cancer mutation database. Moreover, the whole clinical phenotype is consistent with a broad hyperactivation of the RAS pathway. Altogether, these data rather suggest a mosaicism than an isolated and independent occurrence of the two mutations. This mutation is present in both mesodermally and ectodermally derived tissues, suggesting that it may have occurred relatively early during development.

Consistently, previous reports support the hypothesis of oncogenic mosaic mutation as responsible for EN. FGFR3 and PIK3CA mutations observed in the abnormal Blaschko's line are systematically absent from the normal skin counterpart; this has for a long time suggested that EN results from somatic mutations occurring in a metameric pattern during development and leading to a mosaicism.2 5 Hernandez et al reported several patients with multiple EN with the same FGFR3 R248C mutation, but without the severe achondroplastic phenotype that is usually associated with R248C germline mutations.18 Of even greater interest, an undoubted mosaic FGFR3 mutation has been reported in a patient showing multiple EN, seizures and mental retardation.6 The higher risk of cancer conferred by mosaic mutations in FGFR3 or PIK3CA is all the more plausible since several cases of young patients harbouring both EN and low-grade bladder cancers have been reported (case report and review by Garcia de Jalon et al19). To our knowledge, rhabdomyosarcoma has been associated with EN in only two patients.7 20 However, as mentioned above, the constitutive activation of the RAS pathway, characteristic of CS, predisposes to both rhabdomyosarcomas and hyperproliferative skin disorders.12 21 Noticeably, our patient shows a clinical phenotype that can be partially related to CS, since growth defect, feeding difficulties and rhabdomyosarcoma are common features of this syndrome.

From a molecular point of view, our case could also be related to a mosaic form of NS or CFC, a proportion of which are characterised by germline KRAS mutations and occasionally associated with EN. However, the germline mutations reported in CFC and NS almost never involve codons 12, 13 or 61 (review by Aoki et al22 and Gelb and Tartaglia23) since only one CFC patient with a KRAS G12S mutation has been reported.24 Analyses of the functional consequences of KRAS variants responsible for NS and CFC demonstrate their clear ability to hyperactivate RAS signalling, that is nevertheless less pronounced than the one mediated by the G12D variant.9 10 Of note, ubiquitous expression of the G12D K-ras mutant in mice leads to early embryonic lethality.25 Altogether, KRAS-activating germline mutations responsible for NS and CFC might remain viable because KRAS activity is lower than that of G12D variant. This report suggests that a widespread KRAS G12D mutation remains compatible with life provided that it is restricted to a specific mosaic pattern.

Segmental neurofibromatosis type 1 is a well-known cancer-prone developmental disorder due to a mosaic NF1 mutation that leads to a RAS hyperactivation in a specific pattern of tissues. Likewise, mosaicism in HRAS has been described in patients with CS.26 This report describes a comparable genetic abnormality in the RAS pathway that results in a new developmental disorder with cancer predisposition. This observation of segmental G12D KRAS mutation highlights the potentially underestimated importance of oncogene mosaic mutations in childhood cancers.