ESPE Abstracts

Introduction: Growth restriction (GR) comprises ~50% of new patient referrals to paediatric endocrine clinics with <20% receiving a clear diagnosis. Enhanced genetic testing and stratification leading to tailored clinical care is a fundamental need. Whole-genome sequencing (WGS) was offered to patients recruited to the 100,000 Genomes Project (100 KGP), leading to new diagnoses in ~25% of all rare disease participants. The analysis pipeline uses disease-specific virtual gene panels, focusing on coding regions/canonical splice sites of a limited panel of genes known to cause a specific phenotype. Our study aimed to identify potential molecular causes of GR in unsolved 100 KGP subjects.

Methods: Of 72,947 participants recruited to the 100 KGP rare disease cohort, 1,996 GR probands were identified, of which 1,602 (80%) were classified as ‘unsolved’. WGS data were interrogated to investigate coding AND intronic variants affecting canonical/non-canonical splice sites in our extensive virtual gene panel (n = 1,154), copy number variation (CNV) and uniparental disomy (UPD). To investigate beyond known genes, burden analysis examined the effects of other rare damaging coding variants to identify new candidate potentially causative growth genes. Eleven maternal effect genes implicated in DNA methylation disturbance in offspring were examined in the maternal genomes of GR cases.

Results: In unsolved probands, 509 previously unreported coding (83 splice site, 41 frameshift, 35 stop gain, 350 missense) and 217 intronic predicted highly pathogenic variants were identified. This included a novel deep intronic variant in HMGA2 predicted to create a cryptic donor site and possible pseudoexon inclusion in a Silver-Russell Syndrome patient. 207 predicted pathogenic de novo CNVs were identified at previously established regions (1q21, 22q11.2) and novel regions. Burden analysis revealed statistically significant associations with genes implicated in growth including KMT2A and SON. NLRP4, NLRP5, NLRP7 and ZAR1 gene variants were identified in maternal genomes (n = 7) and predicted to affect the DNA methylation statuses of their offspring, causing the GR phenotype.

Conclusions: Childhood GR is multifactorial, with unidentified genetic aetiology in a significant proportion of patients. The discovery of new candidate genes and genomic regions is critical for the identification of new diagnoses/therapeutic targets. We demonstrate that potential molecular genetic diagnoses can be overlooked by standard, semi-automated pipelines. Our novel comprehensive, bioinformatic strategy can enhance the identification of molecular genetic causes in undiagnosed GR phenotypes and could be incorporated into standard care. Furthermore, our methodology can be applied to other endocrine diseases with heterogenous molecular origins.