Predictive Modeling for Clinical Features Associated With Neurofibromatosis Type 1

Cognition and behavior in neurofibromatosis type 1: report and perspective from the Cognition and Behavior in NF1 (CABIN) Task Force

Individuals with neurofibromatosis type 1 (NF1) are prone to the evolution of neurodevelopmental symptomatology including motor delays, learning disabilities, autism, and attention deficits. Caused by heterozygous germline mutations in the NF1 gene, this monogenic condition offers unique opportunities to study the genetic etiologies for neurodevelopmental disorders and the mechanisms that underlie their formation. Although numerous small animal models have been generated to elucidate the causes of these alterations, there is little consensus on how to align preclinical observations with clinical outcomes, harmonize findings across species, and consolidate these insights to chart a cohesive path forward. Capitalizing on expertise from clinicians; human, animal, and cellular model research scientists; and bioinformatics researchers, the first Cognition and Behavior in NF1 (CABIN) meeting was convened at the Banbury Center of Cold Spring Harbor Laboratory in October 2024. This Perspective summarizes the state of our understanding and a proposed plan for future investigation and exploration to improve the quality of life of those with NF1.

Evaluating dimensionality reduction of comorbidities for predictive modeling in individuals with neurofibromatosis type 1

Objective

Dimensionality reduction techniques aim to enhance the performance of machine learning (ML) models by reducing noise and mitigating overfitting. We sought to compare the effect of different dimensionality reduction methods for comorbidity features extracted from electronic health records (EHRs) on the performance of ML models for predicting the development of various sub-phenotypes in children with Neurofibromatosis type 1 (NF1).

Materials and Methods

EHR-derived data from pediatric subjects with a confirmed clinical diagnosis of NF1 were used to create 10 unique comorbidities code-derived feature sets by incorporating dimensionality reduction techniques using raw International Classification of Diseases codes, Clinical Classifications Software Refined, and Phecode mapping schemes. We compared the performance of logistic regression, XGBoost, and random forest models utilizing each feature set.

Results

XGBoost-based predictive models were most successful at predicting NF1 sub-phenotypes. Overall, features based on domain knowledge-informed mapping schema performed better than unsupervised feature reduction methods. High-level features exhibited the worst performance across models and outcomes, suggesting excessive information loss with over-aggregation of features.

Discussion

Model performance is significantly impacted by dimensionality reduction techniques and varies by specific ML algorithm and outcome being predicted. Automated methods using existing knowledge and ontology databases can effectively aggregate features extracted from EHRs.

Conclusion

Dimensionality reduction through feature aggregation can enhance the performance of ML models, particularly in high-dimensional datasets with small sample sizes, commonly found in EHRs health applications. However, if not carefully optimized, it can lead to information loss and data oversimplification, potentially adversely affecting model performance.

Predicting neurodevelopmental disorders using machine learning models and electronic health records - status of the field

Machine learning (ML) is increasingly used to identify patterns that could predict neurodevelopmental disorders (NDDs), such as autism spectrum disorder (ASD) and attention-deficit hyperactivity disorder (ADHD). One key source of multilevel data for ML prediction models includes population-based registers and electronic health records. These can contain rich information on individual and familial medical histories and socio-demographics. This review summarizes studies published between 2010-2022 that used ML algorithms to develop predictive models for NDDs using population-based registers and electronic health records. A literature search identified 1191 articles, of which 32 were retained. Of these, 47% developed ASD prediction models and 25% ADHD models. Classical ML methods were used in 82% of studies and in particular tree-based prediction models performed well. The sensitivity of the models was lower than 75% for most studies, while the area under the curve (AUC) was greater than 75%. The most important predictors were patient and familial medical history and sociodemographic factors. Using private in-house datasets makes comparing and validating model generalizability across studies difficult. The ML model development and reporting guidelines were adopted only in a few recently reported studies. More work is needed to harness the power of data for detecting NDDs early.

The management of neurofibromatosis type 1 (NF1) in children and adolescents

INTRODUCTION

Neurofibromatosis type 1 (NF1) is a rare neurogenetic disorder characterized by multiple organ system involvement and a predisposition to benign and malignant tumor development. With revised NF1 clinical criteria and the availability of germline genetic testing, there is now an opportunity to render an early diagnosis, expedite medical surveillance, and initiate treatment in a prompt and targeted manner.

AREAS COVERED

The authors review the spectrum of medical problems associated with NF1, focusing specifically on children and young adults. The age-dependent appearance of NF1-associated features is highlighted, and the currently accepted medical treatments are discussed. Additionally, future directions for optimizing the care of this unique population of children are outlined.

EXPERT OPINION

The appearance of NF1-related medical problems is age dependent, requiring surveillance for those features most likely to occur at any given age during childhood. As such, we advocate a life stage-focused screening approach beginning in infancy and continuing through the transition to adult care. With early detection, it becomes possible to promptly institute therapies and reduce patient morbidity. Importantly, with continued advancement in our understanding of disease pathogenesis, future improvements in the care of children with NF1 might incorporate improved risk assessments and more personalized molecularly targeted treatments.

The role of scoliosis on the comorbidity and demographics of neurofibromatosis type 1 patients: A retrospective analysis of the National Inpatient Sample database

Neurofibromatosis type 1 (NF1) is the most common neurocutaneous syndrome in the United States, affecting every 1 in 3000 individuals. NF1 occurs due to non-functional mutations in the NF1 gene, which expresses neurofibromin, a protein involved in tumour suppression. As a result, NF1 typically presents with non-cancerous neoplasm masses called neurofibromas across the body. Out of all NF1 abnormalities, the most common skeletal abnormality seen in around 10%-30% of NF1 patients is scoliosis, an improver curvature of the spine. However, there is a lack of research on the effects of scoliosis on demographics and morbidities of NF1 patients. We performed a national analysis to investigate the complex relationship between NF1 and scoliosis on patients' demographics and comorbidities. We conducted a retrospective cross-sectional analysis of the 2017 US National Inpatient Sample database using univariable Chi-square analysis and multivariable binary logistic regression analysis to determine the interplay of NF1 and scoliosis on patients' demographics and comorbidities. Our query resulted in 4635 total NF1 patients, of which 475 (10.25%) had scoliosis and 4160 (89.75%) did not. Demographic analysis showed that NF1 patients with scoliosis were typically younger, female and white compared to NF1 patients without scoliosis. Comorbidity analysis showed that NF1 patients with scoliosis were more likely to develop malignant brain neoplasms, epilepsy, hydrocephalus, pigmentation disorders, hypothyroidism, diabetes with chronic complications and coagulopathy disorders. NF1 patients with scoliosis were less likely to develop congestive heart failure, pulmonary circulation disease, peripheral vascular disease, paralysis, chronic pulmonary disease, lymphoma and psychosis. NF1 patients with scoliosis were predominantly younger, female, white patients. The presence of scoliosis in NF1 patients increases the risks for certain brain neoplasms and disorders but serves a protective effect against some pulmonary and cardiac complications.

Perinatal folate levels do not influence tumor latency or multiplicity in a model of NF1 associated plexiform-like neurofibromas

OBJECTIVE

In epidemiological and experimental research, high folic acid intake has been demonstrated to accelerate tumor development among populations with genetic and/or molecular susceptibility to cancer. Neurofibromatosis type 1 (NF1) is a common autosomal dominant disorder predisposing affected individuals to tumorigenesis, including benign plexiform neurofibromas; however, understanding of factors associated with tumor risk in NF1 patients is limited. Therefore, we investigated whether pregestational folic acid intake modified plexiform-like peripheral nerve sheath tumor risk in a transgenic NF1 murine model.

RESULTS

We observed no significant differences in overall survival according to folate group. Relative to controls (180 days), median survival did not statistically differ in deficient (174 days, P = 0.56) or supplemented (177 days, P = 0.13) folate groups. Dietary folate intake was positively associated with RBC folate levels at weaning, (P = 0.023, 0.0096, and 0.0006 for deficient vs. control, control vs. supplemented, and deficient vs. supplemented groups, respectively). Dorsal root ganglia (DRG), brachial plexi, and sciatic nerves were assessed according to folate group. Mice in the folate deficient group had significantly more enlarged DRG relative to controls (P = 0.044), but no other groups statistically differed. No significant differences for brachial plexi or sciatic nerve enlargement were observed according to folate status.

Neurofibromatosis Type 1-Associated Optic Pathway Gliomas: Current Challenges and Future Prospects

Optic pathway glioma (OPG) occurs in as many as one-fifth of individuals with the neurofibromatosis type 1 (NF1) cancer predisposition syndrome. Generally considered low-grade and slow growing, many children with NF1-OPGs remain asymptomatic. However, due to their location within the optic pathway, ~20-30% of those harboring NF1-OPGs will experience symptoms, including progressive vision loss, proptosis, diplopia, and precocious puberty. While treatment with conventional chemotherapy is largely effective at attenuating tumor growth, it is not clear whether there is much long-term recovery of visual function. Additionally, because these tumors predominantly affect young children, there are unique challenges to NF1-OPG diagnosis, monitoring, and longitudinal management. Over the past two decades, the employment of authenticated genetically engineered Nf1-OPG mouse models have provided key insights into the function of the NF1 protein, neurofibromin, as well as the molecular and cellular pathways that contribute to optic gliomagenesis. Findings from these studies have resulted in the identification of new molecular targets whose inhibition blocks murine Nf1-OPG growth in preclinical studies. Some of these promising compounds have now entered into early clinical trials. Future research focused on defining the determinants that underlie optic glioma initiation, expansion, and tumor-induced optic nerve injury will pave the way to personalized risk assessment strategies, improved tumor monitoring, and optimized treatment plans for children with NF1-OPG.