High-risk phenotypes of genetic disease in a Neonatal Intensive Care Unit population
Genetic diseases are a significant contributor to infant mortality and morbidity, accounting for 35% of deaths during the first year of life. Early diagnosis and intervention are crucial, as more than 25% of neonates with genetic diagnoses can be cured if identified promptly. However, the diagnostic process for genetic diseases in neonates is often lengthy and costly, particularly for those without typical phenotypes such as special facial features or multiple congenital anomalies (MCAs). This study aims to identify high-risk phenotypes indicative of genetic diseases in a Neonatal Intensive Care Unit (NICU) population, using an innovative Auto-Neo-HPO pipeline to analyze electronic medical record (EMR) data.
The study was conducted in a large tertiary NICU at the Children’s Hospital of Fudan University, Shanghai, China, from June 1, 2016, to June 30, 2020. The inclusion criteria were neonates with a postnatal age of less than 28 days, gestational age above 35 weeks, hospital stay for at least 24 hours, and informed consent provided by biological parents or guardians. Neonates with MCAs or missing/low-quality clinical information were excluded. The study utilized EMR data and clinical exome sequencing data to assess demographic characteristics, gestational age, birth weight, clinical phenotypes at discharge, and outcomes.
The Auto-Neo-HPO pipeline was developed to extract and normalize clinical findings from EMR data using the Human Phenotype Ontology (HPO). This tool includes a local semantic database in Chinese and English and a natural language processing pipeline to convert non-database-matched phrases into HPO terms. Two experienced geneticists and two neonatologists reviewed and updated the Auto-Neo-HPO output to ensure accuracy. The pipeline’s performance was validated, and HPO terms were modified to account for neonatal-specific phenotype conditions.
A total of 2,600 neonates were enrolled, including 248 (9.5%) with genetic diagnoses. The study population comprised 1,554 (59.8%) males and 1,046 (40.2%) females. Among the enrolled neonates, 168 (6.5%) died in the hospital or received palliative care, including 33 (13.3%) with genetic diagnoses and 135 (5.7%) without genetic diagnoses. The most common phenotypes in the NICU population were abnormal heart morphology (49.2%), jaundice (47.3%), and sepsis (42.3%).
The study identified three high-risk phenotypes associated with genetic diagnoses: muscular hypotonia, seizure, and cryptorchidism. These phenotypes were significantly overrepresented in neonates with genetic diagnoses compared to those without. Specifically, muscular hypotonia had an adjusted odds ratio (OR) of 3.41, seizure had an adjusted OR of 2.47, and cryptorchidism had an adjusted OR of 3.36. In contrast, jaundice and meningitis were identified as low-risk phenotypes for genetic diagnoses, with adjusted ORs of 0.57 and 0.17, respectively.
The abnormality of metabolism/homeostasis was not significantly different between neonates with and without genetic diagnoses, suggesting that metabolic phenotypes alone may not be reliable indicators of genetic disease in this population. However, specific metabolic conditions such as hyperinsulinemia, glutaric aciduria, lactic acidosis, and hyperlipidemia were strongly associated with genetic diagnoses based on clinical experience, although they were counted less than 10 times in the study and merged into parental HPO terms.
The study also addressed challenges in mapping clinical terms to HPO terms. For instance, different clinical terms such as jaundice, neonatal hyperbilirubinemia, and hyperbilirubinemia were mapped to distinct HPO terms but were merged into a single HPO term based on neonatologists’ recommendations. Similarly, various types of congenital heart disease (CHD) were merged into the HPO term “Abnormal heart morphology,” except for patent ductus arteriosus (PDA), which was analyzed independently.
Several limitations were noted in the study. First, the severity of clinical findings was not described in the HPO terms, potentially limiting the depth of phenotypic analysis. Second, not all clinical findings could be mapped to HPO terms, necessitating the use of similar or parental HPO terms based on the HPO’s topological structure. Finally, the age of onset for each phenotype was not considered, which may be crucial for clinical decision-making.
In conclusion, the Auto-Neo-HPO pipeline effectively identified high-risk phenotypes for genetic diagnoses in a large NICU population. Muscular hypotonia, seizure, and cryptorchidism were identified as significant indicators for further genetic testing. These findings provide valuable insights for clinicians in the early diagnosis and management of genetic diseases in neonates, particularly in resource-limited settings where genetic testing may not be readily available.
doi.org/10.1097/CM9.0000000000001959
Was this helpful?
0 / 0