Paraspinal ganglioneuroma in the proband of a large family with mild cutaneous manifestations of NF1, carrying a deep NF1 intronic mutation

To the Editor : Neurofibromatosis type I (NF1, MIM 162200) is an autosomal dominant neurocutaneous-skeletal syndrome affecting approximately 1 in 3000 individuals worldwide caused by a germline mutation in the NF1 tumour suppressor gene, spanning approximately 280 kb of genomic DNA comprising 57 constitutive and 3 alternatively spliced exons (chromosome 17q11.2) (1). Neurofibromin, the NF1 gene product, is ubiquitously expressed and negatively regulates the ras signalling pathway (2). By puberty, over 90% of patients have developed café-au-lait macules (CALMs), skinfold freckling, neurofibromas and Lisch nodules. NF1-associated tumours include neurofibromas, malignant peripheral nerve sheath tumours, optic gliomas, pheochromocytomas, rhabdomyosarcomas, and juvenile myelomonocytic leukemia. Less than 60 cases of ganglioneuromas, embryonal tumours of migrating neuroectodermal cells, have been published, 9 being spinal in origin (Table S1). Although benign tumours, a resultant mass effect on adjacent vital organs may cause morbidity. Widespread NF1-associated spinal ganglioneuromas have been reported in only one patient carrying an in-frame deletion within the NF1 GAP-related domain, i.e. NM_000267.3:c.4312_4314delGAA; p.Glu1438del (3). We describe a large NF1 family (Table 1, Fig. 1) ascertained after a spinal ganglioneuroma was identified in the 4-year-old female proband. Comprehensive RNA-based NF1 mutation analysis revealed a 80bp inserted sequence derived from NM_001042492.2 intron 31(23a), caused by a novel deep intronic mutation creating an intronic splice acceptor site, used together with a cryptic intronic splice donor sequence located downstream: NM_001042492.2:c.4173+278A > G. This mutation caused out of frame missplicing, but affecting only the transcript containing the alternatively spliced exon 31 (isoform 2): NM_0010424 92.2:r.4173_4174ins4173+279_4173+358 (Fig. 2a–d). Only nine deep intronic NF1 splice mutations were reported so far (Table S2). The proband (IV:4) presented with a paraspinal opacity after chest radiography to confirm resolution of pneumonia. Magnetic resonance imaging (MRI) showed a right-sided paravertebral mass extending from T7–T11. Further evaluation revealed > 6 CALMs but no neurofibromas, skinfold freckling, Lisch nodules or long bone dysplasia. A flat nasal bridge, bilateral epicanthic folds and low set, posteriorly rotated ears reminiscent of a Noonan phenotype were present. Left optic nerve thickening without evidence of an optic glioma was seen on MRI. The paraspinal mass was excised. Pathological and histological examination revealed large numbers of scattered and nested ganglion cells within a background of abundant mesenchymal type stroma, comprising Schwann cells interlaced between bundles of collagen, consistent with a mature ganglioneuroma (Fig. S1). The extended family was clinically and genetically examined (Table 1). A multi-step NF1 mutation approach using an RNA-based core assay, complemented with additional gDNA-based methods provides full characterization of pathogenic mutations in over 95% of classic nonfounder patients meeting NIH criteria for NF1 and is especially powerful for the detection of unusual splice mutations (4). Splicing errors in NF1 are frequent (∼27%) with half of them residing outside the canonically 100% conserved AG/GT dinucleotides (4, 5). This poses significant limitations to gDNA-based NF1 mutation analyses. A nomenclature for the different types of splice mutations was previously proposed, with deep intronic mutations named type II according to this nomenclature (5). This classification may need further refinement based on the current identification of the first deep intronic splice mutation specifically affecting only one of the transcripts. The proband’s deep intronic splice mutation was previously missed in another lab performing clinical gDNA-based sequencing, leading to diagnostic confusion and repeat testing which could have been avoided. In this family, none of the patients presented with cutaneous neurofibromas or externally visible plexiform neurofibromas. NF1 is notorious for its interand intrafamilial phenotypic variability and a consistent mild cutaneous phenotype within a family,