Saving Lydia: A Path from Diagnosis to Treatment in 3 months

Rohan Seth
Lydian Accelerator
Published in
15 min readOct 30, 2019


My 10 month old daughter, Lydia, has a genetic disease that will leave her severely mentally and physically disabled. We are working on a genetic treatment that can change her trajectory. You can read our backstory in Saving Lydia: Why I’m Open Sourcing My Baby.

Since we shared our story, dozens of parents contacted us for help in their own diagnostic or treatment odysseys. While our path is unique to our mutation and still has many unknowns, we wanted to share our first few months in case it helps others. For a much more thorough guide, we highly recommend Matt Might’s An Algorithm for Precision Medicine.

January— The Diagnostic Odyssey

Birth — Lydia was born on her due date, December 22, at University of California, San Francisco. Our pregnancy was healthy; all pre-natal genetic testing was clear. Jen felt abnormal shaking in-utero a few times a day, but those concerns were dismissed.

Seizures — Lydia started having seizures on Day 1. Doctors suspected a brain injury, but an MRI was clear. All bacterial and viral tests came clean too. Finally, we were told “It’ll resolve itself. Go home”. When we insisted on genetic and metabolic testing, they pushed back saying the paperwork was complicated and not worth the effort. At our persistence, they agreed to send out Lydia’s blood for testing 20 common genes associated with neonatal seizures. During this whole time, Lydia had been heavily sedated under a strong anti-epileptic medication called Phenobarbital — she couldn’t breathe or eat on her own. I spent the days reading academic papers to understand what tests to advocate for and Jen pumped every 3 hours so Lydia could be fed through the tube. We left the hospital 3 weeks later once she was able to feed and breathe on her own.

Genetic Test Result — A couple of days later, we got a call from our neurologist. He said he had good news: Lydia had a benign mutation in the gene KCNQ2. This caused early seizures but now that the seizures were in control, she would develop just fine. By now, we had requested hospital records (which were off-limits to us for some reason while we were still admitted). We had read that her background EEG was abnormal — our research suggested that meant something was quite wrong. The genetic report itself said the mutation was “Likely Pathogenic” so it wasn’t clear how the doctor could conclude that it was benign. After questioning his logic for a few minutes, we got the answer we have now learned to dread — “Trust me, I’m an expert!”. Despite Lydia’s continued abnormal movements, he refused to see her. It was our first realization that we were alone in our diagnostic journey.

Finding the prognosis—We spent the next few days combing through every paper and database we could find. There was no existing information on Lydia’s exact mutation. We learned that the amino acid substitution caused by the mutation could theoretically have a pathogenic affect on the protein structure. We learned that the gene was dominant so we should ask the hospital for parental genetic testing; if Mom or Dad also carried the mutation, it could indicate the mutation was likely benign since we were healthy. Shockingly, a standard test like this wasn’t automatically offered to us and we had to advocate for it. A few weeks later, we found out that we didn’t carry the mutation; it had happened spontaneously in Lydia.

Finding other patients— The best signal for her prognosis was finding other patients with the exact same mutation. We found a Facebook group started by parents of other children with KCNQ2. After combing through 4 years of posts, we found two parents who had written about children with the same mutation. One family was in the UK and the other one in Greece. After speaking to them, we found out that their children were older and could not sit, stand or talk. They were non-verbal and severely mentally disabled. We took the information to our neurologists and asked them to brainstorm how we could confirm whether Lydia’s mutation would lead to the same outcome. We got no constructive collaboration from them, we were told to “wait and watch”.

Genotype Phenotype Relationship — “Wait and watch” did not resonate with us. We needed to know if the same mutation (genotype) led to the same outcome (phenotype) in all patients. The Greek family graciously provided some of their medical records and we hired someone to translate them. Lydia’s EEGs were consistent with theirs. Our neurologist had told us that because Lydia’s seizures were controlled earlier, she wouldn’t be affected. We spent the next few days talking to every KCNQ2 expert from around the world — they all agreed that early control of seizures had no impact on the developmental outcome. We also learned that the outcome was closely linked to the mutation site; there was a strong genotype-phenotype relationship. We combed through the Facebook group to find counterexamples, but none were satisfactory.

We never got an “official” diagnosis, but sometime around Lydia’s one month birthday, January 22, we concluded that she was going to be severely affected if we didn’t do anything.


Advocate for your child. Verify everything your doctors tell you. The clinician controlled diagnostic process is broken and archaic. Keep talking to experts till you find someone who is collaborative and genuinely curious. If we had gone with the “wait and watch” approach, we would be months behind in Lydia’s treatment. We don’t know how much more irreversible damage there would be.

Demand genetic testing. Do as thorough testing as you can get (exome, genome) since it can help with future therapeutic targets. It is absolutely ridiculous for doctors to push back against these. These are much cheaper now and to be honest, I don’t even understand why these need to be clinician ordered if you are willing to pay out of pocket. Make sure to do trio testing (father and mother too) if there is a chance of inheritance.

Understand the prognosis of the mutation. Find people with the same or similar mutations using Clinvar and patient communities/registries. Look through Matt’s guide for more advice here.

February — The Treatment Odyssey Begins

We knew there was no available treatment for KCNQ2 yet. In February, we set up a consultation with a neurologist and genetic counsellor at Stanford University and had a remarkably different experience from UCSF. While they reinforced the lack of current options, they brainstormed ways to get us more answers. We were told that there are ways to study a mutation in cell models or model organisms to understand its effect. We were told about Matt Might, another Computer Scientist father of a child with a genetic mutation. Matt had changed career paths, and had built algorithmic models to identify drugs to target genetic diseases. For the first time, we realized that we as motivated parents could do something.

Becoming an “Expert”— First, we had to brush up on basic science. We read about how DNA was transcribed into RNA which is the code to create proteins. DNA/RNA were the software, proteins were the hardware. Traditional medicines (called small molecules) target proteins which made them more complex and harder to engineer. Newer treatments (gene therapy, gene editing, RNA therapies) target the source code. We started pulling on every thread to learn our options. In Computer Science terms, we started with a breadth-first search. We read over 200 academic papers. We talked to pharma executives, hosted KCNQ2 researchers in our living room, traveled to Boston and Chicago. We talked to a black market gene therapist. Our story even made its way to George Church who offered to speak to us.

Functional impact of MutationWe learned that understanding the functional impact of the mutation would be critical to devise treatment options. Most KCNQ2 mutations were “Loss of Function” (LoF) — they prevented the potassium channel gate in the brain from functioning or reduced its function, leading to a partial or complete loss of current. However, some were the opposite. Within the LoF mutations, if the mutation led to complete loss of function in just one copy of the gene, children would have seizures (due to haploinsufficiency) that would resolve themselves and there were no developmental consequences. But, if the mutation not only had no function of its own but also interfered with normal copy, this appeared as loss of activity in both copies, then it was considered dominant negative, leading in a severe outcome. We were lucky to find a collaborator at Northwestern who was studying these mutations, and offered to study Lydia’s mutation. A few weeks later we learned that Lydia’s mutation not only had no function of its own but also interfered with the normal copy through a dominant negative effect.

The search for a drug — We searched for drugs that could improve development in addition to seizure control. Some studies claimed Vitamin B6 could help and given the lack of downside, we started that. We read a bit about the Keto diet. Finally, we learned about Retigabine, a drug that was a potassium channel opener — an exact match for Lydia’s disease-causing mutation. Unfortunately, the medicine had been discontinued a few years ago by the manufacturer GlaxoSmithKline, because they were unable to monetize it. Fuck them, but honestly, I get it — the current pharma industry is modeled around a slow expensive process to create new drugs, so it only makes sense if the drug is useful to large cohorts. This is simply guaranteed to orphan genetic diseases that are all slightly different from each other.

Anyway, a company from Vancouver called Xenon Pharmaceutical had bought the rights to bring Retigabine back to the market. We contacted Xenon through employees, board members and their patient advocacy channel, but eventually learned they would not be able to provide anything to us until the end of the year. A year was too long, we needed it now. We contacted over 50 labs who had used Retigabine in their research to see if they had any left — no luck. Finally, in mid-February, we reached out to a friend who owned a pharma company in India to see if they could manufacture a small batch for us. Over the next few weeks we devised a strategy to buy research-grade material, verify it’s purity against a reference standard and self-compound it at home on a weekly basis. We started dosing Lydia in May and have noticed several improvements since then. While this supply is limited, we hope this can buy us time till the Xenon drug is ready.

But, what about “Gene Therapy” —Retigabine was still just a bandaid; we needed to fix the root cause. We’d heard about gene therapy, but had no clue what it was. Turns out, we had bucketed a few different things in one group — gene editing, gene therapy and RNA therapy. They each had their own advantages and disadvantages, all seemed years away from helping Lydia. On February 14, we were sent a link to Mila’s story, a 7-year-old girl who had been treated with the first genetically customized drug called Antisense Oligonucleotide. This research was completed within 12 months by Dr. Tim Yu at Boston Children’s Hospital. I spoke to Julia Vitarello, Mila’s mom, the next day on the phone and flew to Boston to meet with her and Tim. Could this work for Lydia?


Become an expert. Find collaborative partners who are still actively involved in the gene. What is the current state of research for this gene? What is the current state of Natural History studies? Is there a registry? Who is doing studies in cell models, mouse models, iPSCs? Are there any clinical trials? Find papers on Google Scholar,, Pubmed. If you hit a paywall, use SciHub.

Understand the functional impact of the mutation. If it’s not already published, this means you will need to partner with an academic collaborator. This is critical to understand what treatment options may work.

Find or start a patient organization. How good are they at fundraising and supporting research? The understandable reality is that as most patient-led foundations get older, they lose some urgency as their kids are less likely to benefit from their own efforts. Figure out if you can join their BoD or accelerate their efforts.

March— A game plan

We continued to follow every lead and started diving deeper. We shared a two page document called Saving Lydia (Our Research Strategy). In it, we outlined a three-phase scientific plan and asked for help. We wrote, “We understand that we have been dealt a tough hand — we are not unreasonable in our expectation, but we continue to be ambitious in our goals”

  • Phase 1 Small Molecules: A possible strategy was to repurpose an existing FDA approved drug that was already a good fit for the mutation. There are ways to test a library of thousands of compounds against a patient’s cells to find a match. Given the existence of Retigabine (and our general skepticism that this approach was too dependent on luck vs. engineering), we decided to skip repurposing and just use Retigabine. This has already had a remarkable effect on Lydia’s phenotype (and we have since proven in her cells that the drug is very effective)
  • Phase 2 Antisense Oligonucleotide (ASO): Antisense Oligonucleotides are a true platform technology that hold a lot of promise for genetic diseases. They are delivered at the RNA level, and as a result need to be dosed every few months. Our ASO strategy would be to knock-out an entire copy of the KCNQ2 gene that carries the mutation. The hypothesis is that this would restore the mutation to the benign form of the disease, since plenty of kids were born with just one copy. There was a group thinking about this approach but applied more generally across the entire KCNQ2 population, but their work was too early, had too many risks and we could not develop a meaningful strategy to accelerate it. We concluded that an N-of-1 (single patient) ASO was the fastest way to help Lydia and de-risk the generalized approach by generating important proof of concept data.
  • Phase 3 Gene Editing/Therapy: Finally, is there a one-time fix? There are two strategies here: either replacing the mutant gene with copies of the good gene (gene therapy), or fixing the actual typo itself (gene editing). An early question we asked every researcher was if the work required to do these was “innovation” or was it simply “hard work”. It would be unwise to pin all your hopes on innovation. We learned that KCNQ2 was a large gene and the existing viral vectors that are used in these treatments were not efficient enough to reach all the cells. There were also possibilities for off-target effects, and the FDA was (rightly) going to ask for longer and more expensive safety studies. We decided that we would start early exploratory work here with something called base editors, but the majority of work was innovation and we could not consider this in our critical path for the next 12 months.

We spoke to hundreds of different companies and researchers. A few:

Drug Repurposing: MediKanren, Modelis, Perlara, Q-State

ASOs: Tim Yu Lab, Q-State, DeepGenomics, Ionis, Wave

Gene Editing/Therapy: Steven Gray Lab, Gao Lab, David Liu Lab, Genedit

April — An N-of-1 ASO

Since we had de-risked Phase 1, April was spent on Phase 2. We talked to potential research and clinical leads and put together a high-level step-by-step plan with an ambitious goal to create an ASO for Lydia within 12 months. Our plan allowed us to start on a few critical pieces of work:

  • Exome Sequencing: We got Lydia’s whole exome sequenced so we could identify additional mutations in her gene that could be ASO targets. This could take up to 4 weeks.
  • Identifying Mechanism of Action: ASOs have different mechanisms. Because we were able to perform a functional study on Lydia’s mutation, we identified that our mechanism of action would be similar to what is being used for Huntinton’s Disease: an allele specific knockdown.
  • Start making patient derived cells (iPSCs): Regardless of what we did, we would want to test the ASO in Lydia’s cells which can take 4–6 months to create. We traveled to Chicago in early April to enroll Lydia in a study to start making these iPSCs (thankfully, Lydia traveled like a champ!).
  • ASO designs: More designs mean more shots at the goal. We started lining up a few independent designers. We also started contacting a few larger companies that had the technology to give us much better designs.


By April, we had confidence in our high level treatment plan. Of course, there are still some scientific risks and questions (what level of knockdown can we achieve, what level of knockdown is necessary for functional rescue, and what age would the treatment be most effective). We spent the next couple of months working out critical logistical details:

  • Finalizing our Research Team to develop the ASO. We graded each team based on their financial, regulatory and scientific/innovation risk. An important consideration was how collaborative and open to sharing their technology they were. Another important consideration was whether they insisted on using proprietary technology — this is where patient and researcher interests are not always aligned. Researchers want to use their own latest and greatest technology, but that creates additional work to prove safety that can significantly delay your case.
  • Finalizing the clinical team. Creating the drug is just one small piece. You still need someone who can administer the treatment. They would need to design a trial, get IRB approval (lawyers!) and apply for FDA approval. Most importantly, they would need to be highly motivated and be strong collaborators with the family.
  • FDA Conversation: Our regulatory path was to ask for compassionate use approval. We were concerned that for some reason or the other, Lydia may be excluded from this process. We got connected to a few senior officials at the FDA and walked them through our step by step plan. We were blown away by how supportive and collaborative they were.
  • Begin work on our Minimum Viable Product (MVP). We worked with our collaborators on a Gantt chart. We outlined the risks at every step and put together a contingency for each risk. With this process, we were able to get a reasonable expectation of time and cost, the distribution of outcomes, the critical dependencies and how the hand-offs between institutions would happen. We set monthly goals and created bi-weekly check-ins to stay on top of the project.
  • Fundraising. This research was expensive and fundraising still falls upon families. We set up a non-profit, put together our mission statement and manifesto, and created a website and video.
  • Scaling this platform. We know that what we’re hoping to do for Lydia needs to scale so that it can benefit millions of other children with genetic diseases. There are 3 things that need to scale — the science, the regulatory framework and the business model. Realistically, there are far too many unknowns here and many people will need to step up and try different approaches. For our non-profit, we are currently focused on scaling the scientific side by pursuing an open source model.


We realize we were lucky to be able to put together a plan this quickly. We benefited from the research efforts of others, particularly the KCNQ2 Cure Foundation. Our mission now is to make sure that others after us move much faster.

We were told the role of parents should be limited to advocacy and fundraising. This was simply wrong. Parents funding research need to be in the driver’s seat of scientific decisions, otherwise you are guaranteed to be taken advantage of. We were fortunate to find incredible advisors who helped us along the way. We had friends like Carl Dambkowski, Jon Levy, Aditya Berlia, Elad Gil, Jessica Rannaud, Dena Westphalen, Ben Dean and Jan Hsi Lui who provided critical early advice. We established relationships with Ethan Perlstein, Shi Yin Foo and Ana Mingorance who had dedicated their careers to rare diseases, with Dione Kobayashi and Eric Marcusson who had years of ASO expertise, and with Jeff Caroll, Matt Might, Sonia Vallabh and Eric Minikel, incredible patient-turned-scientists who not only inspired us but spoke our language. There were dozens who responded to cold-emails and hopped on calls (and a few who were unhelpful and self-serving — we learned to eventually screen those folks out).

Lydia spent these first 3 months in extreme pain and discomfort. She was in/out of doctor visits and therapies. She is 10 months old now, much happier but with global developmental delay. Her ASO is still months away (with many unknowns still) and our journey will certainly not end there; our quest to improve her quality of life will continue for the rest of her and our lives. Our road ahead is uncertain, but at this moment in time, we are filled with gratitude to all our advisors, donors and collaborators who have given us this opportunity to create hope. Thank you!

If you would like to support our efforts, please go to:



Rohan Seth
Lydian Accelerator