The child in question, a four-and-a-half-year-old boy, carried a de novo missense variant in the gene SETD1B, also known as KMT2G. This gene encodes a catalytic subunit of the COMPASS complex, a molecular machine responsible for depositing trimethyl marks on histone H3 at lysine 4, or H3K4me3, at gene promoters. These marks are associated with open chromatin and active transcription, so the presence of a loss-of-function variant disrupts this process and leads to chromatin compaction and transcriptional repression at key regulatory loci. The result is a cascade of developmental problems: SETD1B-related syndrome is characterised by developmental delay, intellectual disability, autism spectrum traits and epilepsy, most commonly absence seizures, in which a child momentarily loses awareness, often dozens of times per day.

In this boy’s case, the clinical picture was severe and worsening. From the age of two and a half, he experienced up to 30 atypical absence seizures per day, marked by behavioural arrest, eyelid fluttering, upward eye deviation and occasional head drops, unsteadiness and falls. His electroencephalogram showed the classic 3 Hz spike-and-wave discharges of childhood absence epilepsy. Standard medications proved ineffective or poorly tolerated: ethosuximide and sodium valproate failed to control the seizures, and by age four he had begun experiencing generalised tonic-clonic seizures lasting up to six minutes, typically during infections. Lamotrigine stopped the convulsive seizures, but trials of levetiracetam, clobazam and phenobarbitone were either ineffective or produced behavioural side effects, and he was still suffering as many as 50 absences a day, along with multiple daily tonic seizures on waking. Trio exome sequencing ultimately identified the SETD1B variant c.5686A>G, p.Lys1896Glu, which was absent from population databases and classified as likely pathogenic under American College of Medical Genetics and Genomics criteria.

Faced with drug-refractory epilepsy, the clinical team introduced a Modified Atkins ketogenic diet, a less restrictive variant of the classic ketogenic diet in which carbohydrate is tightly limited while fat supplies most of the body’s energy. Carbohydrate was initially restricted to 15 grams per day, with fat providing 65 to 70 percent of total energy intake and protein allowed to appetite. Remarkably, despite initial food refusal associated with elevated blood ketones of 6.8 mmol/L, the boy became seizure-free during the first week of diet initiation. After a brief transition through a low glycemic index diet to ensure safe ketosis, his macronutrient composition settled at roughly 10 grams of carbohydrate per day, 140 grams of fat and 96 grams of protein, with twice-daily blood ketone monitoring confirming levels within the ideal therapeutic range of 2 to 5 mmol/L. Critically, all antiseizure medications were held constant throughout the three-month trial period, meaning the dietary intervention was the only variable that changed.

The clinical response was dramatic and durable. Within the first month on the diet, absence seizures fell from around 50 per day to between zero and three per day, a reduction of more than 90 percent, and no further generalised tonic-clonic seizures occurred. Even during infections, when seizure thresholds typically drop, his daily seizure count rose only to around 10, compared with more than 50 previously. The benefits extended beyond seizure control: his parents and treating clinicians observed improvements in behaviour, language, memory—as measured by delayed story recall—and concentration, including puzzle completion. Formal measures confirmed these gains. His score on the PedsQL Epilepsy Module, an epilepsy-specific quality of life instrument, improved from 50 at baseline to 61 after three months, and his Clinical Global Impression score improved from 7, indicating severe illness, to 2, indicating marked improvement. Eighteen months after starting the diet, supplemented with medium-chain triglyceride oil, he maintained the greater than 90 percent seizure reduction, with blood ketones optimised at 4 to 5 mmol/L and only one convulsive seizure in that entire period.

What elevates this case report from an encouraging anecdote to a genuine scientific advance is the molecular detective work that accompanied the treatment. The researchers performed single-cell RNA sequencing on 25,159 peripheral mononuclear cells across three samples: the patient before the diet, the patient after three months on the diet, and an age- and sex-matched healthy control child on a standard Western diet. Using the Seurat analysis framework, they identified nine distinct immune cell types, including monocytes, neutrophils, B cells, CD4-positive and CD8-positive T cells and natural killer cells, and then compared gene expression patterns between the patient and the control at each time point, defining differentially expressed genes at a false discovery rate below 0.05 and subjecting them to Gene Ontology over-representation analysis.

The baseline findings revealed just how broadly a single chromatin gene variant can disturb the transcriptome. Before the diet, the boy’s immune cells showed dysregulation of chromatin, ribosomal, immune and mitochondrial pathways. Ribosomal and translational pathways, along with immune and inflammatory programs, were predominantly upregulated, while mitochondrial function and transcriptional and epigenetic processes were largely downregulated. The number of differentially expressed genes per cell type ranged from 59 to 1,129, with neutrophils and CD4-positive T cells showing the greatest burden of change, and with more genes downregulated than upregulated. Among the most significant genes elevated before the diet were FKBP5, a well-known stress-response mediator, along with SMAP2, ZBTB16 and IL1R2, an interleukin-1 receptor decoy that flags inflammatory activation.

After three months of ketogenic therapy, the picture shifted in the opposite direction, and the changes suggested transcriptional normalisation rather than over-correction. In the post-diet comparison, differentially expressed gene counts ranged from 13 to 2,510 per cell type, now dominated by upregulated genes, and the previously elevated stress and inflammatory genes such as FKBP5 and IL1R2 fell back toward the levels seen in the healthy control child. In neutrophils, the most downregulated pathway before the diet, “response to virus,” driven by RNA helicases of the DDX family, interferon-stimulated genes such as RSAD2 and the IFIT family, and interferon signalling mediators including JAK1, was significantly restored after the diet. Conversely, the overactive bacterial defence response pathways quietened. In CD4-positive T cells, the most downregulated pathway, “nuclear speck,” involving genes essential for spliceosome assembly, chromatin modification and stress response, such as SF3B1, WAC and BCLAF1, was upregulated after therapy, while the previously hyperactive cytosolic ribosome genes of the RPS and RPL families were brought back down. The most statistically significant changes were extraordinary in magnitude, with adjusted p-values in neutrophils reaching below 1.29 × 10⁻²⁷³ for FKBP5 and 2.39 × 10⁻²⁴² for IL1R2.

The mechanistic explanation for these effects rests on an intriguing biochemical property of ketone bodies. When the body shifts from glucose to fat metabolism, it produces ketones such as beta-hydroxybutyrate, and these molecules are known to act as inhibitors of histone deacetylases, the enzymes that remove acetyl groups from histones and promote chromatin compaction. By inhibiting histone deacetylases, ketones promote histone acetylation, open up chromatin and enhance transcriptional accessibility. In a disorder like SETD1B-related epilepsy, where a faulty methyltransferase leaves promoters in an abnormally closed state, this HDAC-inhibiting activity may partially counterbalance the transcriptional dysregulation caused by the genetic defect. The researchers note that this is mechanistically intuitive given SETD1B’s role in establishing open, transcriptionally active chromatin, and that the post-diet restoration of nuclear speck pathways supports a genuine improvement in epigenetic regulation and RNA processing rather than a mere pharmacological suppression of seizures.

The findings also echo a growing literature on ketogenic therapy in other Mendelian disorders of the epigenetic machinery. The same research group has previously shown that a ketogenic diet modified ribosomal protein dysregulation in KMT2D-related Kabuki syndrome, where treatment was associated with resolution of episodes of cognitive regression. Aberrant ribosomal protein expression appears to be a shared feature of histone lysine methyltransferase disorders, and the normalisation of translational activity after the diet in the present case suggests that the diet may restore chromatin-mediated control of protein synthesis, supporting broader recovery of transcriptional and epigenetic homeostasis. Taken together, these observations hint that ketosis might represent a broadly applicable metabolic lever for a family of rare chromatin disorders that currently have few targeted treatments.

The authors are careful to acknowledge the limitations inherent in a single-patient study. There was only one patient and one control, no duplicate or longitudinal samples to assess intra-individual variability, and the analysis was performed on peripheral blood cells rather than neurons, since brain tissue is inaccessible in living children. However, they point out that SETD1B is highly expressed in immune cells, making peripheral blood a reasonable surrogate model of the gene’s dysfunction. They propose that future studies incorporate single-cell proteomics and chromatin accessibility assays such as ATAC-Seq and ChIP-Seq in larger cohorts, which would allow researchers to measure directly whether ketone-induced histone modifications are occurring at the promoters of the genes whose expression normalised after treatment.

Even with those caveats, the case stands as a compelling proof of principle for precision medicine in rare disease. It shows that single-cell transcriptomics, even at a scale of one patient, can reveal the molecular fingerprints of a chromatin disorder, document a therapy’s mechanism of action and potentially serve as a biomarker for therapeutic monitoring. It also adds to the evidence that the ketogenic diet, long valued as a pragmatic but mechanistically opaque epilepsy treatment, may exert its antiseizure effects partly through gene regulation. For children with SETD1B-related epilepsy and related chromatin disorders—conditions for which medications often fail and no cure exists—the idea that a carefully monitored diet can nudge a dysregulated genome back toward equilibrium offers a rare note of hope, and a clear roadmap for the larger studies that must now follow.