Abstract
Background: Neurofibromatosis 1 (NF1) is one of the most common inherited disorders characterized by mutations in the tumour suppressor gene NF1. Its clinical manifestations are highly variable and unpredictable. A specific NF1 mutation does not predict the severity or complications of the disease.Objective: The objective of this study was to build an empirical classification scheme without any a priori hypotheses to identify the underlying NF1 subtypes that best explain the observed heterogeneity.Methods: We performed latent class analysis (LCA) of 1351 consecutive NF1 patients aged >17 years seen between 2002 and 2014. Data and phenotypic features were collected prospectively on a standardized form.Results: The median age was 36.8 (17-81) years. A three-class model showed the best fit: 706 (52%) belonged to the LC1 'Cutaneous neurofibromas' class having preferentially cutaneous neurofibromas (99%), plexiform neurofibromas (63%) and blue-red macules (29%); 593 (44%) belonged to the LC2 'Subcutaneous neurofibromas' class characterized by the presence of at least 10 subcutaneous neurofibromas (21%) and a familial form (77%) and 52 (4%) belonged to the LC3 'Dysmorphic phenotype' class characterized by dysmorphic features (78%) and learning difficulties (87%). Patients in LC1 had a higher likelihood of developing scoliosis (RR = 1.7, 95% confidence interval (CI) [1.2-2.4]). Patients in LC2 were more likely to be men (RR = 1.4, 95% CI [1.1-1.7]). Patients in LC3 were at higher risk of having an optic pathway glioma (RR = 4.8, 95% CI [1.9-11.8]) and epilepsy (RR = 4.5, 95% CI [1.8-11.6]).Conclusion: Our findings invite the performance of a larger cohort study to test whether the various latent classes reflect different underlying genetic modifiers of these phenotypic traits.