Timothy Yu and the lab that built the first individualized drug

When Julia Vitarello reached Timothy Yu's lab in early 2017 with whole-exome sequencing data on her daughter Mila that had identified one CLN7 Batten disease mutation but not the second, Yu was 38. He had been an assistant professor at Boston Children's Hospital for four years, a neurogeneticist running a small lab focused on whole-genome sequencing in pediatric neurological disease. The decision to take Mila's case, deepen the sequencing to look for the missing variant, and then design an antisense oligonucleotide to address the variant Yu's team eventually found turned his lab into the first proof of concept for patient-customized antisense oligonucleotide therapy.

That was milasen, published in the New England Journal of Medicine in October 2019. Three further programs from Yu's lab have been published or publicly described since: atipeksen for Ipek Kuzu's ataxia-telangiectasia variant, valeriasen for Valeria's KCNT1-related epilepsy, and a fourth program in development for a different family. The lab is now the first institution other researchers reference when they describe the n-of-1 antisense oligonucleotide model.

Training

Yu's biography is the standard physician-scientist trajectory with one unusual feature, which is the depth of his early commitment to genomics in clinical neurology. He completed his undergraduate degree at Harvard College in biochemistry and molecular biology, then his M.D. and Ph.D. in neuroscience at the University of California, San Francisco. Clinical neurology training followed at Massachusetts General Hospital and Brigham and Women's Hospital. A neurodevelopmental genetics fellowship at MGH and Boston Children's Hospital came after that.

He joined the faculty at Boston Children's in 2010 as an instructor, was promoted to assistant professor in 2013, and is now an attending physician in the Division of Genetics and Genomics at Boston Children's and on the faculty at Harvard Medical School. The career arc is conventional. The work that came out of it is not.

The Yu Lab

The lab Yu runs at Boston Children's is small by industry standards and directionally focused. Its core capability is whole-genome sequencing for the workup of pediatric neurological disease, with a particular emphasis on cases where standard exome sequencing has not produced a diagnosis. The lab's pipeline includes the deeper sequencing that found the SVA retrotransposon insertion in Mila Makovec's MFSD8 gene, the analytical tools to recognize that the insertion introduced a cryptic splice site, and the molecular-biology pipeline to design antisense oligonucleotides against new targets.

The lab's published work establishes a pattern. A family identifies a mutation through standard genetic workup, the workup either fails to find a second variant in a recessive disease or finds a variant whose consequence is unclear, and Yu's lab takes the case forward with deeper sequencing and mechanistic interpretation. When the mutation turns out to disrupt splicing in a way an antisense oligonucleotide can correct, the program proceeds to ASO design, in vitro validation in patient-derived cells, manufacturing under good manufacturing practices through an external contract partner, rodent toxicology, and submission of an emergency or sponsor-investigator investigational new drug application to the FDA.

The interval from family contact to first dose has been roughly 10 to 12 months across Yu's published programs. The cost per program, in published commentary, has been approximately $1.6 million.

What the four cases have established

Milasen (Mila Makovec, CLN7 Batten disease, 2018). The first individualized ASO for a single mutation. Published in NEJM in October 2019. Mila received the drug intrathecally beginning January 31, 2018, at age seven. Her seizure frequency declined and the seizures she did have became shorter. The disease continued to progress. She died on February 11, 2021, at age 10. The case proved that a drug could be designed and manufactured for one mutation, on a timeline of months, under FDA review.

Atipeksen (Ipek Kuzu, ataxia-telangiectasia, 2020). The second program from the lab and the first to demonstrate that the milasen approach was repeatable. Atipeksen targets a splicing defect specific to Ipek's ATM variant. Intrathecal dosing began in early 2020 when Ipek was two years and nine months old, escalating from 3.5 mg to 42 mg over ten weeks. The program has continued, with Ipek receiving doses on a roughly quarterly schedule. She was six years old at the time of recent published commentary, and the lab has reported reductions in disease-progression markers.

Valeriasen (Valeria, KCNT1-related epilepsy, 2020). The third case, and the one that produced the most difficult published outcome. Valeriasen suppressed Valeria's pathogenic KCNT1 transcript in vitro and entered clinical use after an eight-to-ten-week toxicology run. Valeria's seizures decreased dramatically after dosing. Both Valeria and a second child treated under a similar protocol developed hydrocephalus, an intrathecal-route adverse event the FDA's clinical guidance on individualized ASOs has since flagged as a class-known risk. Valeria died of disease progression. The case generated some of the safety data the field now references when designing intrathecal-route programs.

The pattern across the three published cases is that the program shape is reproducible, the chemistry generalizes, and the outcomes are honest. Two of the four named children have died of their disease. One is alive and on continued treatment. The fourth program is in earlier stages.

What Yu has said the lab has learned

In published interviews and at the Oligonucleotide Therapeutics Society and other meetings, Yu has been explicit about what the work has shown and what it has not.

The case selection problem is real. Not every mutation is amenable to splice-switching ASO correction, and not every family that contacts the lab can be helped. Yu's team triages applications by molecular mechanism: splice-disrupting mutations, intronic insertions creating cryptic splice sites, and a small number of other categories are tractable; missense mutations producing misfolded proteins are usually outside the playbook unless they happen to disrupt splicing.

The expectations problem is also real. Families sometimes reach the lab with a child whose disease is too advanced for an ASO to reverse the existing damage, even if the drug works as designed. Mila's case is the clearest example. Yu has said that if the same case were presented today with two more years of pre-symptomatic treatment available, the outcome might have been different.

The data-generation argument is the one Yu has emphasized most consistently in public talks. Each program generates data that the next program uses: safety information for the chemistry class and the intrathecal route, manufacturing process improvements, regulatory precedents, and biomarker measurements. The early cases bear higher risk and produce data that benefits later cases. The case that does not save the child still produces something.

The institutional shape

Boston Children's Hospital has the infrastructure that makes Yu's work possible at the speed at which it works. The hospital has a Division of Genetics and Genomics with the sequencing core. It has a neurology clinic that follows pediatric cases longitudinally and can detect the right candidates. It has institutional review board and regulatory affairs experience with sponsor-investigator INDs. It has the relationships with external GMP manufacturing partners that the milasen, atipeksen, and valeriasen programs depended on.

Other institutions are beginning to replicate the model. Stanley Crooke's n-Lorem Foundation, founded in 2020, has built a parallel philanthropic-and-academic infrastructure that has now treated 50 nano-rare patients with no ASO-related serious adverse events. Programs at Children's Hospital of Philadelphia, the University of Pennsylvania, and Stanford have begun to publish individualized ASO cases. The Yu lab is no longer the only institution doing this kind of work, which is the point. The first lab built the infrastructure. The second and third labs adapted it. The tenth lab will adjust it.

The work that started with one family reaching the lab in 2017 has become the closest the field has to a repeatable program shape for n-of-1 antisense oligonucleotide therapy. The lab still takes a small number of cases per year. The constraint, in 2026, is no longer whether the chemistry works.