Clinical and Mutational Spectrum of Neurofibromatosis Type 1–like Syndrome FREE

Neurofibromatosis type 1 (NF1) (Online Mendelian Inheritance in Man [OMIM] 162200), an autosomal dominant disorder affecting approximately 1 in 3000 individuals worldwide,^1- 2 results from inactivating mutations in the NF1 tumor suppressor encoding neurofibromin. Neurofibromatosis type 1 is characterized by multiple café au lait macules (CALMs), skin-fold freckling, iris Lisch nodules, and tumors of the nervous system such as optic pathway gliomas, astrocytomas, neurofibromas, and malignant peripheral nerve sheath tumors. Other frequently observed features are specific bone abnormalities, short stature, macrocephaly, and learning problems.

The NF1 diagnostic criteria were established at the National Institutes of Health (NIH) NF consensus conference in 1988,³ were updated in 1997,⁴ and are widely used to make the diagnosis using information obtained from physical examination, family history, and radiologic studies.

Neurofibromin is a negative regulator of RAS and pathogenic NF1 (NM_000267.2) mutations result in increased active RAS (guanosine triphosphate–bound RAS) and increased signaling through the downstream effectors such as the mitogen-activated protein kinase (MAPK) pathway.⁵ Recently, a genetically distinct but phenotypically similar disorder caused by heterozygous SPRED1 (NM_152594.2) mutations was identified (OMIM 611431).^6- 8 Sprouty-related EVH1 domain–containing protein 1 (SPRED1), a member of the SPROUTY/SPRED family, negatively regulates MAPK signaling by inhibition of RAS-RAF interaction.^9- 11 Inactivating SPRED1 mutations, resulting in increased MAPK signaling, were found in individuals from 5 families with multiple CALMs, freckling, and macrocephaly, with many fulfilling the NIH criteria for NF1 based on the presence of CALMs, freckling, or presence of a positive family history.⁶

In the original study,⁶ families suitably sized for linkage analysis had specifically been collected based on individuals with multiple CALMs with or without freckling and absence of a detectable NF1 mutation, which may imply that the observed phenotype and possible complications associated with the disorder might have been biased or underestimated. Similarly, the study by Spurlock et al⁸ and Pasmant et al⁷ primarily focused on SPRED1 analysis in patients with the mild pigmentary phenotype. Given the absence of neurofibromas in any patient described so far,^6- 8 it has been proposed to refer to this new syndrome as Legius syndrome (as named in OMIM 611431)^7- 8 and discontinue referring to it as neurofibromatosis type 1–like syndrome (NFLS).

In this study, we determined the phenotype in 22 unrelated probands and 18 relatives carrying a SPRED1 loss-of-function (LOF) mutation identified through clinical genetic testing. In addition, we carried out a cross-sectional study in an anonymous cohort of 1318 unrelated patients referred for NF1 genetic testing, in whom no NF1 mutation was found, allowing delineation of the SPRED1 mutational spectrum and estimation of the frequency of NFLS.

The flow of participants included in the studies is shown in Figure 1.

In a cross-sectional study, a SPRED1 LOF mutation was identified in 26 NF1-negative probands (with clinical NF1 testing before July 2007) for whom a blood sample was received (August 2007–September 2008) for clinical SPRED1 mutation testing (this test became available by August 2007 and is described at http://www.genetics.uab.edu/medgenomics). Probands were referred from 19 different centers in the United States and Canada. Written informed consent to collect phenotypic and genotypic data was obtained from study participants and parents or guardians of children. Clinical notes and phenotypic information as available through the referring physician were reviewed.

Height and head circumference at a given age were converted to percentiles using the Centers for Disease Control and Prevention US growth charts (http://www.cdc.gov/nchs/data/ad/ad314.pdf). An individual was recorded as macrocephalic when head circumference was above the 97th percentile. An individual was recorded to have relative macrocephaly when head circumference was above the 97th percentile at the age when height would have been at the 50th percentile. Information on race/ethnicity was recorded by the referring health care clinician as African American, Asian, Hispanic, Native American, Caucasian, other (eg, Pacific Islander) or “not provided.” Additional detail on ethnic background was requested if essential for interpretation of the genetic results. The study was approved by the University of Alabama at Birmingham (UAB) Institutional Review Board.

Samples received at the UAB Medical Genomics Laboratory for NF1 genetic testing (August 2003–July 2007) were included in an anonymous study if a phenotypic checklist (http://www.genetics.uab.edu/medgenomics) summarizing the NF1-related clinical signs at the time of blood sampling had been completed by the referring physician. This is a cross-sectional, referral-based study with samples received from more than 250 different referral centers in the United States (85%), Canada (8%), Latin America (2%), Australia and New Zealand (2%), Asia (1.5%), and Europe (1.5%). This inevitably implies heterogeneity in the quality of the data set, which is based on these phenotypic data forms. In total, 2432 unrelated samples with phenotypic data were available by July 2007, which equals approximately 85% of samples received.

Search for genotype-phenotype correlations using a Microsoft Access (Microsoft, Redmond, Washington) database containing phenotypic and mutation data was approved by the UAB Institutional Review Board. The following data were entered through the password-protected Access database: year of birth, age of the individual at time of completion of the phenotypic checklist, country of origin, sex, race/ethnicity, sporadic or familial, standardized information pertaining to the NF1-related signs, and a unique identifier not linked to date of birth, name, or other database connected to such information. The same unique identifier was labeled on an aliquot of the leftover sample from the patients referred for NF1 clinical testing. Entries were exported to an Microsoft Excel table format to facilitate the subsequent statistical analyses.

In 1114 of 2432 individuals, an NF1 mutation had been identified. Anonymous DNA samples from 1318 NF1-negative probands underwent further research-based SPRED1 mutation analysis, resulting in 3 groups of phenotypic data: NF1 mutation–positive, SPRED1 LOF mutation–positive, and NF1 and SPRED1 LOF mutation–negative cohorts. This NF1-negative cohort is not biased toward the initially reported phenotype in patients with a SPRED1 mutation.⁶ Genetic research on the anonymous samples was approved by the UAB Institutional Review Board. The 26 probands in the clinical cohort are most likely also part of the anonymous cohort because these patients previously had an NF1-negative test result during the study period.

Comprehensive NF1 and SPRED1 mutational testing was performed as described in the eAppendix and eTable 1. The pathogenicity of identified missense mutations was analyzed using 2 different functional assays: (1) neurite outgrowth of pheochromocytoma 12 (PC12) cells in vitro after stimulation with nerve growth factor and (2) Elk-1 activation and phosphorylation. Both functional assays are described in detail in the eAppendix. Secondary structure prediction of mutant p.Thr102Arg and p.Cys74Phe SPRED1 was carried out as described in the eAppendix.

Ninety-five percent confidence intervals (CIs) of proportions were calculated with the exact binomial method using Statcalc version 1.1 (http://www.ucs.louisiana.edu/~kxk4695/StatCalc.htm). We also calculated sensitivity, specificity, and positive and negative predictive values for detection of a SPRED1 mutation in the NF1-negative cohort, also using Statcalc.

In the clinical cohort, a SPRED1 mutation was identified in 27 NF1-negative probands (eTable 2); 22 had a SPRED1 LOF mutation (including 2 LOF missense mutations: p.Thr102Arg and p.Pro415Ala); 4 of 6 missense mutations and 1 silent mutation were classified as rare, probably benign variants: p.Cys74Arg, p.Ser149Asn, p.Asp398Asn, p.Cys433Tyr, and c.42T>C (eTable 3). Forty-three relatives of the 22 probands with a SPRED1 LOF mutation participated in the study and 18 of 43 carried the family-specific SPRED1 mutation. Six of 22 were de novo cases (27%; 95% CI, 11%-50%). Phenotypic details of 40 individuals with a SPRED1 LOF mutation are provided in Table 1 (aggregated data) and eTable 2 (individual data); 14 of 40 had more than 5 CALMs and axillary/inguinal freckling. An additional 6 individuals fulfill NF1 criteria if family history is taken into account. The majority of individuals (35/40) had more than 5 CALMs even at a young age. Axillary/inguinal freckling was mild or faint in 7 of 14 patients with freckling (Figure 2A).

Noonan syndrome (OMIM 163950; an autosomal dominant disorder characterized by a variable combination of facial dysmorphism, short stature, pectus deformity, and congenital heart defects¹²) was suspected in 1 child (individual S10-III1; individual/family numbering is shown in eTable 2), a mild pulmonic valve stenosis was present in individual S11-I2, and mild to severe pectus excavatum was seen in 3 individuals (Figure 2B and eTable 2). Five children had abnormal language and speech development and 5 were reported to have attention deficit, hyperactivity, or both. None of the individuals carrying a SPRED1 mutation showed neurofibromas, typical osseous lesions, or a symptomatic optic pathway glioma.

We observed 1 occurrence each of tenosynovial giant cell tumor, angiolipoma, breast cancer, and dermoid tumor of the ovary and 2 lipomas in 1 individual (eTable 2). No tumor material was available to study the causal involvement of SPRED1.

In family S23, the NF1 mutation c.2755delG had previously been identified in a classically affected patient, S23-III1 (eTable 2), but no NF1 mutation was found in several of his relatives with NF1 signs (individuals S23-II2, S23-II3, S23-I2, and S23-III3). A truncating SPRED1 mutation was found in individual S23-II3 (the younger brother of individual S23-II2), in individual S23-I2 (the mother of individual S23-II2 and S23-II3), and in individual S23-III3 (the child of individual S23-II3) but was absent in individuals S23-II2 and S23-III1.

Spectrum of SPRED1 Mutations. We identified 34 different SPRED1 mutations in 43 probands in the anonymous cohort of 1318 unrelated NF1-negative patients (eFigure 1). The majority of mutations (23/34) are predicted to result in a premature stop codon. Furthermore, 1 in-frame deletion (c.242_256del15bp; p.Ile81_Val85del), 9 different missense mutations, and 1 silent mutation were found. Mutations were spread over the entire gene; 23 are novel. Eight different mutations (p.Arg16X; p.Arg18X, p.Arg24X, p.Arg64X, p.Arg117X, p.Thr313Met, p.Arg325X, and p.Asp398Asn), accounting for 12 of the 43 mutations, can be attributed to deamination at methylated CpG dinucleotides. No SPRED1 copy number changes were detected.

SPRED1 Missense Mutation. The phenotype of individuals carrying a (probably) benign missense mutation is described in eTable 3.

Functional and In Silico Analysis of Missense Mutations. SPRED1 missense mutations p.Thr102Arg and p.Pro415Ala are pathogenic and mutants p.Cys74Phe, p.Ser149Asn, p.Met188Ile, p.Thr196Ile, p.Thr313Met, p.Asp398Asn, and p.Cys433Tyr are probably benign variants. Data supporting classification of missense mutations, including the functional effect on the RAS-MAPK pathway using the PC12 differentiation assay and Elk-1 reporter assay and secondary structure prediction by in silico homology modeling, are summarized in the eAppendix, eTable 4, and eFigure 2, eFigure 3, and eFigure 4.

Phenotypic Characteristics. We identified 33 LOF SPRED1 mutations (eFigure 1) in an anonymous NF1-negative cohort of 1318 unrelated probands presenting with a broad range of NF1-related signs, not biased toward the phenotype initially described in patients with a SPRED1 mutation.⁶ Demographic and phenotypic features of a cohort with and without SPRED1 mutations were compared with a cohort of 1114 patients with a definitive NF1 mutation (Table 2 and Table 3). Mean age was 14.9 years in the NF1-positive cohort, 9.5 years in the SPRED1-positive cohort, and 12.4 years in the NF1/SPRED1-negative cohort. The average number of NIH criteria fulfilled was 2.39 in the NF1-positive cohort, 1.84 in the SPRED1-positive cohort, and 0.89 in the NF1/SPRED1-negative cohort.

The high proportion of sporadic patients likely reflects the greater uptake of genetic testing among patients with diagnostic uncertainties in the absence of a family history. Thirty-one of 33 SPRED1 LOF–positive individuals had more than 5 CALMs with or without freckling and no other NF1 criteria, 13 (39%; 95% CI, 23%-58%) of which were sporadic cases.

The ratio of SPRED1/NF1 mutations detected in the total group of 2432 patients was 33/1114 (3.0%; 95% CI, 2.0%-4.1%): 20/251 (8.0%; 95% CI, 4.9%-12.0%) in the familial group vs 13/863 (1.5%; 95% CI, 0.8%-2.6%) in the sporadic group.

Of 1086 patients fulfilling NIH criteria for clinical diagnosis of NF1, an NF1 mutation was found in 823 (76%; 95% CI, 73%-78%), a SPRED1 mutation in 21 (1.9%; 95% CI, 1.2%-2.9%), and no NF1/SPRED1 mutation in 243 (22%; 95% CI, 20%-25%), with 211 of 243 being sporadic cases.

No SPRED1 mutations were found in any NF1-negative patients with neurofibromas, optic pathway glioma, Lisch nodules, or typical osseous lesions. The sensitivity, specificity, and positive and negative predictive values to detect a SPRED1 mutation in the analyzed NF1-negative cohort are reported in eTable 5. The highest positive predictive value was observed in familial patients with more than 5 CALMs with or without freckling and no other criteria (0.720; 95% CI, 0.506-0.879), with a sensitivity of 0.545 (95% CI, 0.364-0.719) and a specificity of 0.995 (95% CI, 0.988-0.998).

We investigated the clinical spectrum of a neurofibromatosis type 1–like syndrome, recently named Legius syndrome (OMIM 611431), and estimated its frequency relative to NF1 in an anonymous cohort of patients with a broad range of signs typically found in NF1. Individuals with SPRED1 mutations presented with multiple CALMs with or without freckling. The dermatologic phenotype in young children with a SPRED1 mutation could not be differentiated from NF1 and half of individuals (20/40) with a SPRED1 mutation fulfilled the NF1 diagnostic criteria based on presence of more than 5 CALMs with or without skinfold freckling and with or without familial history.

Most patients presented with a mild phenotype compared with NF1, although in family S24, severe progressive dystonia, a large temporal venous anomaly, and a vascular anomaly were present. In the report by Spurlock et al,⁸ 1 patient had an inguinal hemangioma, which is of interest given the recently identified association of RASA1 with abnormal angiogenesis. RASA1 (p120-RASGAP; OMIM 139150), another guanosine triphosphate–ase–activating protein, down-regulates the RAS-MAPK pathway similarly to neurofibromin.^13- 15 The potential association of NFLS with vascular malformations warrants further investigation. The RAS-MAPK pathway syndromes show a large variability in tumor predisposition, congenital malformations, and intellectual disabilities.¹⁵

Because SPRED1 is involved in regulation of the MAPK pathway, patients with SPRED1 mutations may be at increased risk of developing specific tumors or learning problems. In our study, 5 children had abnormal language and speech development and 5 were reported to have attention deficit/hyperactivity. The importance of SPRED1 for hippocampal-dependent learning was recently documented in mice.¹⁶ Future studies need to systematically investigate potential problems with speech, learning, and behavior in children with SPRED1 mutations.

None of the individuals in our study had neurofibromas, typical osseous lesions, or symptomatic optic pathway gliomas. In this and previous studies, no systematic occurrence of any tumor type was observed, except possibly subcutaneous lipomas in adults, also frequently observed in the general population. Tenosynovial giant cell tumor, dermoid tumor of the ovary, Wilms tumor, tubular colon adenoma, acute leukemia, and non–small cell lung cancer were each observed in only 1 individual once.

Combining data on 53 adults with NFLS from this study and in previous reports^6- 8 results in a post hoc power estimate of 80% to detect complications occurring in at least 5% of affected adults. Using the binomial distribution, the chance of detecting 2 cases with a specific complication with a prevalence of 5% in a group of 53 individuals is 76%. Combining data from this and previous reports^6- 8 on individuals aged 5 years or older with a SPRED1 mutation (n = 79), we estimated, in post hoc power calculation, that this sample size would allow detecting complications with a prevalence of at least 3% with a power estimate of 80%. None of these 79 individuals had evidence of plexiform neurofibroma, discrete neurofibroma, or symptomatic optic pathway glioma, suggesting that the frequency of these complications is lower than in NF1 (plexiform neurofibromas, 24%; discrete neurofibromas, 53%; symptomatic optic pathway gliomas, 4.3%¹⁷).

However, even these pooled data are underpowered to detect rare complications associated with a prevalence of only 1%; a study of 250 well-characterized, preferably adult patients would be needed. Other limitations of the present study are the referral bias (mainly diagnostic uncertainty) and heterogeneity of clinical data (many clinical NF experts involved).

As of July 2007, no NF1 mutation had been identified through clinical testing in 1318 of 2432 unrelated patients who fulfilled inclusion criteria for further anonymous studies at the UAB Medical Genomics Laboratory. Many samples were submitted because of a diagnostic uncertainty: 55% fulfilled fewer than 2 NIH criteria and 87% were sporadic. SPRED1 analysis of this anonymous NF1-negative cohort revealed a SPRED1 LOF mutation in 33 of 1318. No SPRED1 copy number changes were found in this or previous studies^6- 8 but we recently identified 3 different multiexon deletions, indicating the need for dosage analysis in the clinical setting (L.M., unpublished data).

Clinical features of the cohort with a SPRED1 LOF mutation (n = 33) were compared with the cohort without mutations in NF1 and SPRED1 (n = 1285) and with the cohort with an NF1 mutation (n = 1114). Twenty-one of 1087 individuals (approximately 2%) fulfilling NIH diagnostic criteria in the entire anonymous cohort carried a SPRED1 mutation. All individuals carrying a pathogenic SPRED1 mutation were found in the subgroup of patients with multiple CALMs with or without freckling but fulfilling no other NIH criterion. No SPRED1 mutations were found if patients had neurofibromas, optic pathway glioma, ophthalmologically proven Lisch nodules, or a typical NF1-associated osseous lesion. The data from this anonymous cohort further underscore that association of these features with NFLS must be rare.

In this study of 957 sporadic patients with CALMs with or without freckling and no other criterion, 414 (43%) had a NF1 mutation in the blood and 13 (1.3%) had a SPRED1 mutation. An NF1 mutation was found in 150 of 159 familial patients (94%) with more than 5 CALMs with or without freckling and an additional criterion, usually the presence of neurofibromas, but in 69 of 94 familial patients (73%) with CALMs with or without freckling only. In this last group, 18 of 94 families (19%) showed a SPRED1 mutation, and in 7 of 94 (7%), no mutation could be identified in NF1 or SPRED1.

The NIH NF1 criteria, originally designed to help identify families suitable for linkage studies that led to the positional cloning of the NF1 gene, are widely used and allow establishment of a clinical diagnosis of NF1 in most cases. In individuals with CALMs with or without freckling and no other specific distinguishing features, presenting sporadically or with a family history of CALMs with or without freckling only, the NIH criteria cannot reliably distinguish NF1 from NFLS. In such patients, a correct diagnosis has important implications for prognosis, counseling, and potential prenatal genetic diagnosis. Although an NF1 diagnosis may become apparent with the passage of time, the diagnosis will remain uncertain for individuals who do not develop other signs of NF1.

Molecular genetic testing can resolve the diagnosis in most such cases. In case of diagnostic uncertainty, we recommend that NF1 should be analyzed first and, if negative, SPRED1 testing should be considered in patients with CALMs with or without freckling and no other NF1 diagnostic features. Identification of a SPRED1 mutation may relieve a psychological burden from families who otherwise would be in a waiting mode for potential serious NF1-associated manifestations.

We currently are conservative regarding the clinical surveillance of SPRED1-positive patients and recommend the same medical follow-up as that for patients with NF1. Less stringent surveillance may possibly be recommended for these patients if clinical data from several hundreds of patients confirm the low frequency of benign and malignant tumors.

For all of these reasons, it is important that clinicians, including general practitioners, clinical geneticists, pediatricians, ophthalmologists, dermatologists, neurologists, and oncologists, who are involved in the care, diagnosis, and treatment of individuals with NF1, should be aware that Legius syndrome can resemble NF1.