OR07-6 Integrating Targeted Bioinformatic Searches of the Electronic Health Records and Genomic Testing Identifies a Molecular Diagnosis in Three Patients with Undiagnosed Short Stature

Background: Short stature is a common reason for referral to a pediatric endocrinologist. Despite adequate evaluation, a pathological diagnosis is not identified in the vast majority of patients. It is often difficult to determine which of these many patients have an undiagnosed genetic cause. The electronic health record (EHR) improves our ability to identify cohorts of patients with a specific phenotype of interest, and it has become a valued tool for the augmentation of clinical characterization to drive discovery of phenotype-genotype associations. The aim of our study was to assess the feasibility of using dense phenotypic information from the EHR to systematically identify a distinct cohort of patients with short stature who have a high probability of harboring a monogenic etiology for their short stature. Methods/Results: As an initial proof of principle, we chose to focus on patients with Insulin-like growth factor I (IGF-I) resistance. We performed a targeted bioinformatics search of the EHR at three leading pediatric endocrine departments (Boston Children’s Hospital (BCH), Children’s Hospital of Philadelphia (CHOP) and Cincinnati Children’s Hospital Medical Center (CCHMC)). Inclusion criteria were height below -2 SD (for age and sex) and an IGF-I level above the 90(th)percentile. Patients with known underlying genetic conditions, other chronic illnesses, or precocious puberty were excluded. All eligible patients were approached by mail or telephone to participate in the study. Whole exome sequencing (WES) was performed on DNA extracted from whole blood or saliva samples from ten patients and their immediate family members. The targeted bioinformatics search identified a total of 234 patients (104 at CCHMC, 64 at CHOP and 62 in BCH). Of these, 39 patients were eligible for recruitment and 10 were successfully recruited for genetic testing. WES identified two novel pathogenic variants in IGF1R including a novel missense variant (p.Val1013Phe) and a maternally inherited single amino acid deletion (p.Thr28del) in two patients. In addition, a third patient was found to have a novel missense variant in CHD2 (p.Val540Phe). Functional analyses confirmed the pathogenicity of the p.Val1013Phe variant, and are ongoing for the other two variants. Conclusion: This novel approach combining bioinformatics and modern genetics successfully led to the identification of a genetic etiology in 3 of 10 patients, giving a yield of 30% including identification of two patients with novel pathogenic mutations in IGFIR. Notably, these patients were otherwise missed clinically as having a genetic condition. Similar algorithms can be designed to identify diverse cohorts of patients likely to have genetic conditions. This approach can be integrated into the EHR to generate best practice advisories to advance our ability to diagnose rare genetic conditions and improve patient care.