The ASO field and especially the individualized ASO field can only move forward when we share information, both the good stories and the disappointing and unexpected findings. Tim Yu and group set a very good example for this and it is one of the mottos of N=1 Collaborative (that Tim co-founded).

I applaud the authors for sharing the work and the data (there is a LOT of data in the publication - have a look yourself). I also really appreciate that they shared the hydrocephalus findings already early on so everyone using ASOs could be alert for this potential side effect.

More work is needed to further elucidate this and I know people are studying valeriasen in cell and animal models with reverse translation experiments (knowing that it causes hydrocephalus, can we see anything in model systems.

Furthermore, it has been reported also for nusinersen treated SMA patients, though there it is very rare. Authors speculate it is likely a combination of a high dose and a specific ASO (e.g. tofersen used in ALS patients does not cause it but is given at a high dose).

It is possible that KCNT1 mutation carriers are more sensitive to developing hydrocephalus. However, it has not been reported in patients or animal models. It could be triggered by the ASO though. Notably, authors mention, hydrocephalus has also been reported in Huntington patients treated with ASOs

Authors discuss that the mechanism for the hydrocephalus is not clear. It could be due to reduced CSF uptake or increased production or both. Also it could be triggered by an inflammatory response, though the ASO did not show an inflammatory profile in preclinical studies.

Hydrocephalus was discovered in an early stage, treatment stopped and a shunt placed to reduce pressure. After the stop of the treatment the seizures returned. After a gap of 2 years, treatment is now started again with a lower dose and seizures have reduced again.

Given the poor condition before the treatment the family opted to stop treatment and the patient was transferred to palliative care, where she passed away 3 months later :(
The second patient was treated later and monitored closely for hydrocephalus. She also developed dystonia.

For the first 10 months all was OK for patient 1, but then she developed dystonia, which was improved after treatment. This could be ASO or disease related. Then the patient turned out to have hydrocephalus. The treatment was stopped and placement of a shunt discussed with the family.

Then the patients were treated with a dose escalation every 2 weeks, followed by a maintenance dose of 40 mg every 8 weeks. Patient 1 was first treated in September 2020 and patient 2 in June 2021. Seizures declined for both patients, for patient 2 already in the dose escalation phase.

Authors tested potential off-target sites for the ASO and there were only very limited numbers. in an RNAseq analysis they showed that only KCNT1 was knocked down, none of the partial off target containing genes and no pathways were disturbed due to the ASO.

The other 2 ASOs caused seizures and behavioral abnormalities and were not studied further. The lead candidate was also tested in a chronic study for 15 weeks with repeated dosing, where it was tolerated well. Histology afterwards showed only minimal findings of pathology.

Authors then filed for an IND with FDA and started to perform safety studies with clinical grade ASOs, 2 lead candidates and 1 backup. Only 1 of the lead candidates was safe, and caused only mild transient side effects (hindlimb paralysis, common finding for IT injected ASOs in rats).

Then authors used their lead candidate in the mouse model as it had perfect homology. This showed reduced Kcnt1 levels and improved channel activity in vivo in brain. Mice were treated with intracerebroventricular injections. Treatments also improved survival.

Note that authors do everything right: they have control ASOs that they use as a reference (ASOs could influence channel activity measurements) and they also use an isogenic control line to compare channel activity (mutant has higher activity, but reduced after ASO, but not control ASO).

Authors screened many ASOs in cultured cells to eventually narrow down to a few ASOs, 1 allele selective and 3 non selective. These were tested also in patient-derived IPSC differentiated neuronal cells where they reduced expression of KCNT1 and improved the increased channel activity.

As such, ASO knockdown is likely safe. Note also that there is an extra safety margin here, because the mutation causes increased channel activity. So even if you reduce also the wild type expression, a little bit of mutant protein is still very active.

Authors describe how they first tested if ASO treatment could improve the phenotype in a Kcnt1 mouse model: indeed here RNase H ASOs reduced Kcnt1 levels and seizures. In humans it is known that loss of function of one allele is well tolerated.

The disease is rare, and authors here developed ASOs for 2 patients with the same severe mutation. Both patients had severe epilepsy and hypotonia and limited motor function. Both patients were treated when they were 3 years of age (i.e when 50% of patients has died already).

About 50% of patients die before the age of 3 years. KCNT1 codes for a sodium activated potassium channel and the mutations cause an increased activity of the channel. As mutations are gain of function, RNase H antisense oligonucleotides (ASO) could be an option.

KCNT1 missense variants can cause severe epilepsy in infants, where seizures start shortly after birth. This epilepsy is refractory (i.e. does not respond to anti-epilepsy treatment) and patients also suffer from developmental delay and intellectual disability.

#apaperaday Today's pick is by Nakayama et al in @naturemedicine.bsky.social the publication about Valeriasen that just appeared. N=1 Collaborative origami paper because of the link to N1C and because the first author is Japanese. DOI: 10.1038/s41591-026-04314-9

Authors conclude they have provided proof of concept for their LNP delivery method for enzymes. Of course a next step would be to test it in disease animal models ideally for multiple enzymes and multiple diseases. I like the clear way authors explain everything and outlining the different steps.

Naked enzymes resulted also in antibodies against the enzymes, while this was much reduced for the LNPs. Note that of course here a human enzyme is delivered to a mouse, the the immune response is expected, but in the LNPs the enzymes are shielded.

There was no increased delivery for liver, and mildly increased delivery in spleen and kidney. Infusion reactions occurred for naked enzymes after repeat delivery in mice, with some mice even dying. This did not happen for the LNP enzymes.

Addition of serum reduced delivery somewhat. Then finally LNPs were injected via intravenous infusion in healthy mice. Compared to naked enzymes, delivery was increased, as measured by enzyme activity, in skeletal muscle and diaphragm but not in heart.

Then the amount of PEG was optimized for the ALC-0315 lipid. The morphology of the LNPs was 100-120 nm, with spherical bilayer vesicles. Authors then tested the LNPs in cultured muscle cells showing better uptake than for naked enzymes as measured by enzyme activity.

LNPs were produced at low pH which was then titrated up to a physiological pH (7.4). Particles were 120 nm in size and authors showed that the enzymes were protected from cleavage, but still enzymatically active after delivery.

Authors show that encapsulation levels increase until 1 mg/kg for GAA and then level out. Authors used DOTAP as a lipid, but this turned out to be toxic at higher doses. Authors then used alternative lipids where ALC-0315 worked well and was not toxic in cell culture

If they use uncharged lipids, there is very little encapsulation. Authors then added PEG (poly ethylene glycol), which prevents renal clearance. PEG levels need to be optimized, as too much PEG also prevents tissue uptake.