Abstracts

Rationale: Angelman syndrome (AS) is a neurodevelopmental disorder caused by deficiency of the maternally inherited Ube3a gene in neurons. Promising disease modifying therapies to reinstate production of UBE3A are under development. In particular, an antisense oligonucleotide (ASO) targeting the Ube3a antisense transcript (Ube3a-ATS) has been used to unsilence the paternal allele and increase production of UBE3A protein in a mouse model of AS. Non-invasive biomarkers to detect target engagement and treatment response are needed to support planned human clinical trials. We have previously shown that delta power (2-4 Hz) measured in the scalp EEG is a reliable biomarker for AS in humans and mouse models that reflects cognitive function. However, in cross-sectional studies, delta power measurements vary widely between individuals and throughout development. A natural history model, incorporating longitudinal measurements of delta power, and accounting for age and elapsed time between measurements, may provide a more sensitive measure of treatment effect.

Methods: We utilized a dataset consisting of 204 longitudinal EEG recordings from 56 AS patients to develop a model of delta power at a second visit, given the delta power at an initial visit, age, and elapsed time between visits, intervisit interval (IVI). We used this natural history model to compute the sample and effects sizes needed to detect a treatment induced deviation in delta power outside of the expected natural history with 80% power. We then fit the model on a longitudinal AS mouse dataset and measure for a treatment effect after ASO-treatment compared to placebo-treatment. Finally, we compare deviations from the natural history model with Ube3a expression in Ube3a-ATS ASO-treated and control ASO-treated mice.

Results: In humans, delta power at a second visit can be reliably predicted using the natural history model: Delta_Visit2~Delta_Visit1+log₁₀(Age_Visit1)+IVI+(1|Subject) where each predictor was found to be significant: Delta_Visit1, β -0.33, 95% CI [0.17-0.49], p< 1e-4; log10(age), β -0.096, 95% CI [-0.215, -0.116], p=0.005; IVI, β -0.016, 95% CI [-0.029, -0.0016], p=0.029. Using this model, treatment effects that result in a 5-10% reduction in delta power can be detected with 80% power from a sample size of 25 subjects. Applying this natural history model to the AS mouse dataset, we detect a treatment effect in each week analyzed from 2 to 8 weeks post-treatment (Figure 1, p≈0). Using the mouse model, we also find that deviations in delta power from the expected natural history positively correlate with Ube3a expression (example in Figure 2, p< 0.001).

Conclusions: Deviations in delta power from a human natural history model in AS can detect ASO-mediated improvement in Ube3a expression in AS mice. Future work to validate the relationship between delta power and UBE3A expression after effective treatment in humans with AS would secure delta power as a mechanistic biomarker to gauge both target engagement and therapeutic response in clinical trials.

Funding: Please list any funding that was received in support of this abstract.: NA.