Recurrent NF1 gene variants and their genotype/phenotype correlations in patients with Neurofibromatosis type I

Abstract Neurofibromatosis type I, a genetic condition due to pathogenic variants in the NF1 gene, is burdened by a high rate of complications, including neoplasms, which increase morbidity and mortality for the disease. We retrospectively re‐evaluated the NF1 gene variants found in the period 2000–2019 and we studied for genotype/phenotype correlations of disease complications and neoplasms 34 variants, which were shared by at least two unrelated families (range 2–11) for a total 141 of probands and 21 relatives affected by Neurofibromatosis type I. Recurrent variants could be ascribed to the most common mutational mechanisms (C to T transition, microsatellite slippage, non‐homologous recombination). In genotype/phenotype correlations, the variants p.Arg440*, p.Tyr489Cys, and p.Arg1947*, together with the gross gene deletions, displayed the highest rates of complications. When considering neoplasms, carriers of variants falling in the extradomain region at the 5′ end of NF1 had a lower age‐related cancer frequency than the rest of the gene sequence, showing a borderline significance (p = 0.045), which was not conserved after correction with covariates. We conclude that (1) hotspots in NF1 occur via different mutational mechanisms, (2) several variants are associated with high rates of complications and cancers, and (3) there is an initial evidence toward a lower cancer risk for carriers of variants in the 5′ end of the NF1 gene although not significant at the multivariate analysis.


| INTRODUCTION
Neurofibromatosis type I (MIM#162200), a genetic condition featured by various combinations of multiple café-au-lait spots, axillary and inguinal freckling, multiple cutaneous and subcutaneous neurofibromas, and iris Lisch nodules, is burdened by a high rate of complications, such as optic pathway gliomas, plexiform neurofibromas, osseous dysplasias, and cancers, which can severely impair the quality of life of the affected patients and finally increase the morbidity and mortality for the disease. 1,2 In Neurofibromatosis type I, malignancies reduce the average life expectancy of the affected patients by [10][11][12][13][14][15] years and represent the main cause of death for the disease. 2 The responsible gene, Neurofibromin 1 (NF1, MIM *613113), an oncosuppressor acting as a negative regulator of the Rat sarcoma (Ras) cascade, is characterized by a high rate of de novo pathogenic variants (about 50% of the affected patients harbor novel alleles) and by the substantial absence of hot spots, which have built up our knowledge of the disease as a combination of individual clinical histories with a high inter-and intra-familial phenotypic variability. 3,4 Our understanding of the genotype/phenotype correlations for Neurofibromatosis type I has also been hampered by technical and clinical issues, which have reduced the number of patients undergoing to the genetic test for the disease. 5,6 In fact, both the large size of the gene, composed of 61 coding exons, and a selective indication to perform the genetic testing have brought to a little use of the genetic information, which is not needed to confirm often obvious phenotypes and is not able to predict the clinical outcome for most of the NF1 genetic variants, with few exceptions, such as large deletions having a severe phenotype and other missense and inframe deletions (p.Arg1809, p. Met1149, and p.Met992del) associated with the absence of cutaneous neurofibromas. [7][8][9][10][11][12] All these reasons have confined the molecular analysis to a minority of the affected patients and to the diagnostic dilemmas, mostly children with a discrete number of café-au-lait (CAL) spots with no other associated feature, with few possibilities to perform correlations for degenerative disorders, like cancer, related to the type and site of NF1 variant. 13 In recent years, instead, the advent of the next-generation sequencing (NGS) and the raising demand for pre-implantation diagnosis have increased the access to the genetic test, which is revealing new aspects of the disease, like the features of the novel variants, 14  The issue of genotype/phenotype correlations in Neurofibromatosis type I is further complicated by the mechanism of action of the NF1 gene, which acts as an oncosuppressor through the combination of a constitutional variant (inherited or de novo) and a somatic, acquired inactivation of the other allele for the initiation of the molecular cascade in most of the affected tissues. 20

| NF1 genetic test
NF1 genetic test has been previously described. 14  Classification of the genetic variants has been performed through the classical 5-tiered system, based on standard criteria. 26

| The database
The recurrent NF1 gene variants were extracted from the diagnostic

| Recurrent variants
The analysis of our records including 1077 gene variants revealed 141 unrelated NF1 affected probands who shared 34 variants, each of them recurring at least 2 times in apparently unrelated families. Table 1 shows their frequency distribution, ranging from 2 (12 variants, 24 probands) to 11 (whole gene deletions, shared by 11 probands), their type, and the demographic features of the patients. All the variants are listed in Table 2 and their distribution along the gene is shown in Figure 1 (Tables 2 and S1). Two variants instead, shared by seven patients, were not associated with any complication: the c.1185 +1G>A (three unrelated patients of 12, 15, and 27 years of age) and the c.4267A>G (p.Lys1423Glu) (four unrelated patients of 1, 2, 3, and 37 years of age). By grouping the recurrent variants according to their localization in the NF1 sequence (for this purpose, the NF1 protein domains were used, see Figure 1), no significant differences (χ 2 9.5, p = 0.2) emerged among the groups for the overall rate of complications (Table 3). When the age-dependent penetrance of complications was analyzed, the known higher frequency of complications for the copy number variants (CNVs) of the NF1 gene was confirmed, when compared with the variants in the extradomain region at the 5 0 end or to the rest of the gene sequence (log-rank = 4.93, p = 0.026 and logrank = 9.97, p = 0.0016, respectively, Figure 2).        Concerning the specific genotype/phenotype correlation, apart from the whole-gene deletions, which show the known high frequency of cognitive impairment (Figure 2), 16 the rate of complications or neoplasms presents a wide range ( Also the data on the p.Arg1947*, which is reported with a milder phenotype, 45 are only partly confirmed, since we have found 2 tumors (one rhabdomyosarcoma at 19 year of age and one hypothalamic astrocytoma at 12) out of the 14 patients harboring the variant.

| NF1 gene variants and cancer
Finally, our data support the association with cardiovascular anomalies of the p.Lys1423Glu variant, 9 which was found in one child with pulmonary valve stenosis (Table 2).
When the variants were grouped according to their NF1 gene region, those located in the 5 0 end preceding the first protein domain In conclusion, our study on recurrent NF1 variants shows that the features of recurrence in the NF1 gene are not univocal but involve the most common mutational mechanisms in humans; furthermore, the genotype/phenotype correlation indicates that several variants are associated with a very high risk of disease complications and suggests an initial link between variants in the 5 0 end of the gene and a lower risk of cancer complications of the disease, although not significant at the multivariate analysis.