Molecular genetic analysis of candidate genes for glutaric aciduria type II in a cohort of patients from Queensland, Australia.

Glutaric aciduria type II (GAII) is a heterogeneous genetic disorder affecting mitochondrial fatty acid, amino acid and choline oxidation. Clinical manifestations vary across the lifespan and onset may occur at any time from the early neonatal period to advanced adulthood. Historically, some patients, in particular those with late onset disease, have experienced significant benefit from riboflavin supplementation. GAII has been considered an autosomal recessive condition caused by pathogenic variants in the gene encoding electron-transfer flavoprotein ubiquinone-oxidoreductase (ETFDH) or in the genes encoding electron-transfer flavoprotein subunits A and B (ETFA and ETFB respectively). Variants in genes involved in riboflavin metabolism have also been reported. However, in some patients, molecular analysis has failed to reveal diagnostic molecular results. In this study, we report the outcome of molecular analysis in 28 Australian patients across the lifespan, 10 paediatric and 18 adult, who had a diagnosis of glutaric aciduria type II based on both clinical and biochemical parameters. Whole genome sequencing was performed on 26 of the patients and two neonatal onset patients had targeted sequencing of candidate genes. The two patients who had targeted sequencing had biallelic pathogenic variants (in ETFA and ETFDH). None of the 26 patients whose whole genome was sequenced had biallelic variants in any of the primary candidate genes. Interestingly, nine of these patients (34.6%) had a monoallelic pathogenic or likely pathogenic variant in a single primary candidate gene and one patient (3.9%) had a monoallelic pathogenic or likely pathogenic variant in two separate genes within the same pathway. The frequencies of the damaging variants within ETFDH and FAD transporter gene SLC25A32 were significantly higher than expected when compared to the corresponding allele frequencies in the general population. The remaining 16 patients (61.5%) had no pathogenic or likely pathogenic variants in the candidate genes. Ten (56%) of the 18 adult patients were taking the selective serotonin reuptake inhibitor antidepressant sertraline, which has been shown to produce a GAII phenotype, and another two adults (11%) were taking a serotonin-norepinephrine reuptake inhibitor antidepressant, venlafaxine or duloxetine, which have a mechanism of action overlapping that of sertraline. Riboflavin deficiency can also mimic both the clinical and biochemical phenotype of GAII. Several patients on these antidepressants showed an initial response to riboflavin but then that response waned. These results suggest that the GAII phenotype can result from a complex interaction between monoallelic variants and the cellular environment. Whole genome or targeted gene panel analysis may not provide a clear molecular diagnosis.

Development of a Novel Ulcerative Colitis Endoscopic Mayo Score Prediction Model Using Machine Learning.

Background and Aims: Endoscopic assessment is a co-primary end point in inflammatory bowel disease registration trials, yet it is subject to inter- and intraobserver variability. We present an original machine learning approach to Endoscopic Mayo Score (eMS) prediction in ulcerative colitis and report the model's performance in differentiating key levels of endoscopic disease activity on full-length procedure videos. Methods: Seven hundred ninety-three full-length videos with centrally-read eMS were obtained from 249 patients with ulcerative colitis, who participated in a phase II trial evaluating mirikizumab (NCT02589665). A video annotation approach was established to extract mucosal features and associated eMS classification labels from each video to be used as inputs for model training. The primary objective of the model was a categorical prediction of inactive vs active endoscopic disease evaluated against 2 independent test sets: a full set with a baseline single human expert read and a consensus subset in which 2 human reads agreed. Results: On the full test set of 147 videos, the model predicted inactive vs active endoscopic disease via the eMS with an area under the curve of 89%, accuracy of 84%, positive predictive value of 80%, and negative predictive value of 85%. In the consensus test set of 94 videos, the model predicted inactive vs active endoscopic disease with an area under the curve of 92%, accuracy of 89%, positive predictive value of 87%, and negative predictive value of 90%. Conclusion: We have demonstrated that this machine learning model supervised by mucosal features and eMS video annotations accurately differentiates key levels of endoscopic disease activity.