Q&A With Parents of UPD/ICD Patients and Chief Science Officer Allyson Berent

In response to thoughtful questions from parents of UPD/ICD patients, Chief Science Officer Allyson Berent wrote out the following. If you or a loved one has specific questions for FAST, please email them to We’ll try to post the answers regularly on our website, and then re-post here for the broadest access to these answers. 

I am speaking to you here as the CSO of FAST. These are only my interpretations, opinions and knowledge based off of the scientific literature, routine conversations with the sponsors working in the AS space, and our incredible research team that is working on figuring all of this out. I am NOT speaking for any specific company and ONLY speaking generically here to help answer questions that absolutely deserve answers.

🧬Q1: When will we see the ASO trials recruiting UPD patients?

  • This is not an answer that we can give you. This is up to each sponsor, and they will have their reasons for including/excluding each of the genotypes. What we do know is that they all seem “willing” to include UPD/ICD/Mutation in addition to deletion (and we are pushing for mosaic of course as well). The questions are: when, why not now, and what do they need to make those decisions so we can help them accelerate to the point of including EVERYONE? We have specifically reached out to ALL players about this question, and ask every year to have answers to this. I want to be clear that I am referring to everyone in the space that we get answers from and not a single company, even though I am now a part-time consultant for Ultragenyx Pharmaceutical.  So far, each company for gene therapy or ASOs has stated that they want to have the most consistent group for the earliest phases of clinical trials (Phase 1/2), and in many cases they want the most severely affected to start with. That is generally considered deletions +/- mutations. From a symptomatic perspective, there is not tremendous differences between UPD/ICD and mutations, and there are many deletion kiddos that have functional scores similar to the non-deletion patients. Generally, from data in the natural history study, the deletion patients seem to score a little lower than the others, but they are broadly most consistent. So, once they can show safety (Phase 1) and early efficacy (Phase 2)—which are generally run together in trials as a Phase 1/2—then they will likely add a cohort of other genotypes. 

🧬Q2: What exactly needs to happen for UPD to be added to Inclusion criteria for the three ongoing ASO trials?

The #1 need: They told us they want safety for the first group of patients to understand if the drug is well tolerated. If that turns out to be the case, which they need time to show (maybe 1-2 years), then they will consider moving to the other groups. For gene therapies, they will want to make sure there is no evidence of UBE3A overexpression, but that will be the same concern for ALL genotypes, as this has nothing to do with paternal gene activation (traditional gene replacement therapy of UBE3A re-instatement, not the CRISPR, miRNA, shRNA, etc that are paternally activating using an AAV for delivery). We know scientifically that ASOs should absolutely not show overexpression in deletions and mutations (non-gain of function mutations), but can potentially give over expression (maximal 200%) in the other genotypes (UPD, ICD, Mosaic [some cells and not all]). So, they do not want to start there, so as to not mix learning the safety of the drug itself versus safety in a population if overexpression occurs. Once they have this safety data, that will be the time to expand to the others.

In addition to just showing the safety in non-UPD/ICD/Mosaic patients there is some scientific rigor that is happening to support the safety concerns of overexpression. Not that all companies are requiring this, but regardless work is being done to support safety. This includes the following:

  • Understanding the electrophysiologic (how they send impulses) and morphologic (how they look) behaviors of AS neurons compared to neurotypical neurons is important. From there, we will be able to understand how different genotypes, like deletion, mutation, UPD, ICD, are different from each other. Then, understanding what happens in those neurons (and organoids) if you activate the paternal copy of the gene. Does it rescue this electrical activity and morphology back to normal? Does it make it look more like an overexpression neuron (Dup15q-like)? Then looking at titrating the drug from low dose to high dose to try and see if there are overexpression issues that can be mitigated by lowering the dose. Remember: This is NOT the same as a real brain. What you see in neurons is a complete exaggeration of what you might see if you deliver a drug into the lumbar spinal cord fluid (CSF) and expect all that drug to get to every neuron of the brain. It doesn’t. It is really hard to get the same amount of knockdown (silencing) in the brain as you get in a dish; some regions will get there, and some will not. So, there is a possibility that some regions will have overexpression and some will have perfect expression, compared to the deletion and mutations patients. This work is being done in a few labs, including North Carolina State University with Albert Keung, Yale University with Yong Hui-Jiang, and the group at UConn, where Stormy Chamberlain used to work before she left and went to Roche. So, people are looking actively to ensure it is well-understood.
  • Next, there is data being established in the UPD mouse model to look and see what happens to the mouse when you give it an ASO and allow for the “double knockdown.” That is being done in two different labs currently as well (Erasmus University in Rotterdam and UC Davis). We will share that data as soon as it is available. 
  • Additionally, there was some great work done (reported in closed sessions at the FAST Translational Research Symposium in 2020 and at this year’s Research Conference at the ASF in Austin) looking at overexpression. It was shown when 2x-4x of UBE3A is expressed, more like a Dup15 model. When this was evaluated the mice with overexpression to 2x (max what you would get with a UPD/ICD having paternal activation) was not as pathogenic as once considered, supporting treated the UPD/ICD genotypes with this approach.
  • Finally, all of the ASO drugs in current clinical trials had go through a tremendous amount of rigor, including studies in monkeys for safety and tolerability. The companies that recently discussed this data (in closed session and those that recently reported them) said that they did not see any side effects consistent with “autism like features” when they caused “over expression” of wild type monkeys. Now, do we know what autism features in monkeys really look like, and how to monitor for that? We likely don’t, but at least there was nothing they reported that supported concerns for this. That will be for the regulatory agencies to decide when adding this population to the studies.

References can be found here and here.  

  • One final point I would like to make for mosaic: More people are hesitant to treat mosaic patients because you are definitely going to cause overexpression in some subset of cells that we know are expressing the maternal allele of UBE3A. But this is a minority of cells (1-20% generally), AND the other 80-99% of cells really will then be similar to ICD as they are generally an ICD mosaic. We have tried to emphasize that this is not more of a concern in this population…no genotype will get left behind! 

🧬Q3: Would it be possible to add UPD participants as part of new cohorts within current phases, or would that be impossible as Exclusion criteria are strictly locked for the duration of the Phase?

  • This is absolutely possible to be added as a separate cohort for these genotypes. This can be done in a phase 1/2, or might be done as a separate phase 1/2. That will be up to the sponsor.  There are different ways sponsors will handle this but it can be handled without a new trial.

🧬Q4: In clinical trials in general, is it possible for Phase 3 to have more genotypes included, or would this require a separate new trial?

  • They can all be included depending on how the company designs the trial. They would likely be looked at in sub-groups for all of the reasons in question 1, but hopefully their inclusion in the phase 1/2 will allow them to understand where to place them for the Phase 3. Analysis of groups with similar ages and genotypes, or other common factors (“sub-groups”) is generally done.

🧬Q5: UPD is excluded from ASO trials right now. Can you see a reason why UPD would be excluded from any other non-ASO trial? We are aware that Neuren's "downstream treatment" trial includes UPD; this question is trying to understand whether UPD would be considered as risky in any way in other "upstream" therapeutics. If there are no additional risks, are you aware of any reason why UPD could still be excluded from the upcoming trials?

  • This is a great question. Down-stream targeted therapies (e.g Neuren) will likely include all genotypes, but just look at them separately in the analysis. This is because these are not knocking down the antisense, so there is no clear presumed safety risk for that genotype to cause overexpression. This is isolated to the paternal activation strategies: ASO, SNA, ATF, CRISPR, shRNA, miRNA. Gene therapy (maternal replacement) or Enzyme Replacement, would not have the same concerns as paternal activation either, since they are replacing what is missing. So, there is not a safety reason to avoid those genotypes. There are different risks there for all genotypes of cellular overexpression. Again, some might still want to start with the most severely affected, but again all kids with AS are severely affected so that always boggles my mind. This is a regulatory thing more than anything it seems, and each sponsor will make these decisions individually.

🧬Q6: How many patients would be expected to participate in Phase 3 of current ASO trials? What defines this number?

  • That is up to the sponsor. They will be different and likely depends on the statistical plan calculated based off what they feel would be most likely the endpoint used as a primary endpoint. Then they must calculate how many patients you would need to prove a statistical difference from placebo/sham. That number determines the number of patients needed. More genotypes will need more patients if you want to look at them all separately. Same with age. The more variables you include, the more patients you need. I would venture to guess well over 100 patients for a phase 3 trial if the population is large enough. In ultrarare diseases they are far less patients, yet diseases like AS (more commonly rare) there are far more patients to accomplish this.

🧬Q7: If the scientists don't think that UPD would lead to double production of UBE3A compared to a Deletion genotype in ASO treatment, then what is the expected level of UBE3A? Is there a model, or does it have to be established in a human trial?

  • I would not say they don’t think that. They actually do think that it could cause 2x the amount of UBE3A than neurotypical. It might not because you don’t really get 100% of protein expression when you knockdown the antisense in vivo (in animal models/brains), and you don’t get 100% knockdown in all neurons. So therefore, if you get double, it is not everywhere. So based off the current data published in papers to date it seems more likely to get 150% than 200%, but not in all regions, some regions could be 110% and other regions 200%. I am speaking only in general terms based off what is published in the literature, and—remember--these are animals like mice, rats and monkeys, NOT humans. We expect potentially even less in humans because as you get into bigger brains, it’s harder to get the biodistribution (penetration of the drug to the deep brain structures) that you see in a smaller brain like a rodent. The monkey CNS compartment is over 10x smaller than that of a human. A rat is 1000x smaller and a mouse is about 3500-4000x smaller. Just to give some perspective.

🧬Q8: What is known about ASO efficiency regarding UBE3A levels in neurons? What % of typical for age/sex is currently achieved?

  • This is going to be drug specific and therefore we can’t answer. In neurotypical neurons we know that the paternal allele can make (“or leak”) about ~8-13% of UBE3A. It is not 100% turned off. There are different thresholds it seems in different parts of the brain as well, so some areas 30% knockdown seems to allow for expression and in other areas you need 80% knockdown to get expression. It’s quite complicated.

🧬Q9: What UPD specific research is being considered by FAST/ASF? We understand that there's a lot that FAST/ASF is pushing that would benefit all genotypes; this question is about UPD specifically. What exactly could UPD organoids and cell lines do to help UPD therapeutics specifically? Some examples would be useful.

  • See question 2…I hope I answered that. In short: creating cell lines (biorepository) and organoids out of the UPD/ICD genotypes (honestly the only line we don’t have 5 patients for is ICD, as we could not get those families to participate compared to all the others. If you are reading this and want to help out, please contact Yale! (here is link to the site, and you can write to them directly here:, neuronal cell line landing pads to understand UPD/ICD/Mosaic snoRNA impact in cell lines and organoids, testing the UPD mouse (some feel has a bad phenotype and can’t be reliably tested; others feel differently), and looking at the overexpression mice like Dup15 inducible for overexpression.

🧬Q10: What does the scientific community want to know about UPD? What are their unanswered questions?

  • They want to know if paternal activation of UPD is dangerous. We are doing all we can to answer that. So far it does not look like it is.

🧬Q11: Does the unavailability of UPD animal models hinder research or trials introduction with regard to UPD genotype? If it does in any way, what could be done about it?

  • There is a model, but most feel it is not a great model. Most companies don’t want to use this model and are comfortable with the data from the cell lines and organoids in order to consider going into patients. That’s what we are being told, over and over again (because I ask over and over again). 

🧬Q12: Is there still some UBE3A produced in neurons in UPD case? If yes, why and how much (% of a healthy level)?

  • In neurotypical neurons we know that the paternal allele can make (“or leak”) about ~8-13% of UBE3A. It is not 100% turned off.

🧬Q13: Why is UPD milder than Deletion in its phenotype? 

  • The deletions are generally missing not just UBE3A (like you might expect with a nonsense point mutation) but they are also missing 8-12+ other genes. They are basically missing 5-6 MILLION base pairs of DNA on the maternal allele. So they are absolutely NULL for UBE3A (minus the leak on the paternal) and they have other genes that are haplo-insufficient, missing from maternal as well, which also contribute to tonic inhibition, seizures, light coloring etc. So together it makes their features a bit more severe. UPD/ICD are basically not missing a single base pair of DNA. How is that for some perspective? Considering how different they are the scales of their disease severity are all within a similar range of severity supporting that the most important gene that is not functioning or missing is UBE3A.

🧬Q14: What is the latest research on newborn screening for AS? Would the screening test be able to detect UPD and ICD?

  • We have collaboratively funded (FAST/ASF/Dup15 and PWS) a NBS pilot in the state of North Carolina through RTI.
  • In addition, we have been lobbying other states to work with them to get on their panels. We are now being added to the NBS in NYC (Columbia University), working with the state of Wisconsin (a national leader in NBS), and Rady’s Children’s in California. Wisconsin and North Carolina will pick up UPD/ICD for sure. The others are adding us to the Whole Genome Sequencing (WGS) panels, so that will likely not pick up UPD/ICD. Once we get to their tier 1 (likely when we make it to phase 3 trials) they will add AS to tier 1 and that will then add methylation testing. All will cover deletion and mutation. Australia is already doing it. We are working with the UK and Europe now to get started as part of the FAST Global Search and Rescue Initiative  ( 

🧬Q15: What is the latest epidemiology consensus on the frequency of UPD and ICD?

  • 3-7% of the AS population which is estimated at 1:10,000-1:20,000. The NBS will help tell us this more accurately and this is the goal of the FAST Global Search and Rescue initiative. The more patients we know with each genotypes the better. PLEASE SIGN UP!!!!! Email us here at: to get involved.