Phenotypic heterogeneity of neurofibromatosis type 1 in a large international registry

Abstract

Neurofibromatosis type 1 (NF1) is a rare genetic disorder, characterized by the development of benign and malignant nerve tumors. Although all individuals with NF1 harbor genetic alterations in the same gene, the clinical manifestations of NF1 are extremely heterogeneous even among individuals who carry identical genetic defects. In order to deepen the understanding of phenotypic manifestations in NF1, we comprehensively characterized the prevalence of 18 phenotypic traits in 2051 adults with NF1 from the Children's Tumor Foundation's NF1 registry. We further investigated the coassociation of traits and found positive correlations between spinal neurofibromas and pain, spinal neurofibromas and scoliosis, spinal neurofibromas and optic gliomas, and optic gliomas and sphenoid wing dysplasia. Furthermore, with increasing numbers of cutaneous neurofibromas, the odds ratio of malignant peripheral nerve sheath tumor increased. Phenotypic clustering revealed 6 phenotypic patient cluster subtypes: mild, freckling predominant, neurofibroma predominant, skeletal predominant, late-onset neural severe, and early-onset neural severe, highlighting potential phenotypic subtypes within NF1. Together, our results support potential shared molecular pathogenesis for certain clinical manifestations and illustrate the utility of disease registries for understanding rare diseases.

Keywords: Dermatology; Genetic diseases; Genetic variation; Neurological disorders; Neuroscience.

Figure 1. Association of clinical traits by Pearson correlation.

An 18 × 18 matrix displays the r value by Pearson correlation of each pairwise clinical trait coassociation, using 2051 patient samples. The shade of each cell corresponds to the r value of the Pearson correlation between the traits on the corresponding row and column. Only associations significant to P < 1.73e4 (after Bonferroni’s correction for multiple comparisons) were plotted. Strong associations include axillary and groin freckling, optic glioma and sphenoid wing dysplasia, spinal neurofibroma and scoliosis, and spinal neurofibroma and pain.

Figure 2. Hierarchical clustering of traits reveals consistency with correlations.

Hierarchical clustering with Ward D method shows how traits with the highest degree of similarity across 2051 patients cluster together.

Figure 3. Clustering reveals phenotypic subtypes visualized by t-stochastic neighbor embedding plot.

K-means clustering was performed on the principal components of all traits for 2051 patients and resulted in 6 distinct clusters of patients, representing subtypes of NF1. Clusters were visualized using t-stochastic neighbor embedding (t-SNE) method of dimensionality reduction to create a 2-dimensional plot. Clusters were color-coded by severity of subtype, with green being the least severe and red being the most severe, and the separation of colors visually represents the distinctness of clusters.