Cerebral palsy may be a symptom cluster, not a single disease

Clinical takeaway: Pathogenic variants identified in children with CP do not necessarily explain the motor phenotype, and further workup may be needed when the gene lacks established CP association. 

Genome sequencing has become a routine part of the diagnostic workup for cerebral palsy (CP), and it regularly returns pathogenic variants in genes reported to cause the condition. But what a positive result means remains harder to interpret. A finding that truly bears on the motor phenotype can end a diagnostic odyssey and redirect management, while one that does not can close a workup that should have continued. The interpretation problem has grown alongside the gene list itself, which now runs to hundreds of candidates drawn from sequencing studies of children with the diagnosis. 

CP was long attributed to perinatal events such as prematurity, infection, and hypoxia-ischemia. But population-based studies show those factors account for a relatively low proportion of cases. The findings from a decade of sequencing have produced no consensus on interpretation. 

Some investigators treat the underlying Mendelian conditions as CP mimics. Others hold that a genetic finding should not change a clinical diagnosis of CP, and still others argue the two diagnoses should coexist. A new analysis approaches the problem from the other direction, treating CP as a phenotypic feature that many different conditions can produce and asking which reported gene associations withstand statistical testing, in the published literature and a newly sequenced pediatric cohort. 

Across 21 published sequencing cohorts, variants deemed causal of CP had been reported in 515 genes. The new analysis found statistically significant evidence of association with CP for only 89 of them. The authors are explicit that the remaining genes are not ruled out. A failed test reflects insufficient evidence, not proof of no association. CP is also underreported in the genetics literature, which makes the 89 likely an undercount. 

"We decided to do this study because there has been no accepted statistical method to determine if a disease-associated gene is truly linked to CP," says Adam Arterbery, first author, of the Jackson Laboratory for Genomic Medicine. "We discovered it was extremely difficult to find statistical evidence of causality for many of the genetic variants previously linked to CP." 

The pattern repeated in the newly sequenced cohort. Whole-genome sequencing of 460 children diagnosed with CP identified pathogenic or likely pathogenic variants in 15.8% of children, involving 60 genes, but only 16 of the 60 had statistically significant evidence of CP association in the literature analysis. Some findings were interpretable as incidental. For example, three boys carried pathogenic G6PD variants, a gene tied to CP only through untreated kernicterus though none had a history of it. 

"We have developed a paradigm that considers CP to be a phenotypic trait rather than a precise disease diagnosis," says Peter Robinson, corresponding author, of the Jackson Laboratory for Genomic Medicine and the Berlin Institute of Health at Charité. "A better understanding of specific genetic variants has the potential to improve our understanding of the biology of the disease." 

Enrichment separated the strongly supported genes cleanly. CTNNB1, the most enriched gene in the literature analysis, appeared in eight of the 21 cohorts and in 55 of 405 individuals reported elsewhere in the genetics literature, while genes such as LIPH rested on a single reported individual. 

Children with an identified variant differed clinically from those without, with higher seizure rates, more term births, and motor and communication function scores shifted toward the severe end of the scales. 

The literature analysis drew on 21 published next-generation sequencing cohorts totaling 5,440 individuals with CP, plus roughly 5,500 additional publications describing nearly 44,000 individuals with variants in the candidate genes, tested for enrichment against the background population prevalence of CP. The new cohort comprised 460 children aged 4 to 17 diagnosed with CP at Shriners Children's facilities in the US and Mexico City, sequenced by whole genome with parents where available. 

Clinical utility and patient benefit need validation. That means prospective studies of whether stratified gene evidence improves outcomes, plus deeper phenotyping than current cohorts carry. For clinicians, the payoff would be concrete. Gene-level evidence could indicate which children with a pathogenic variant warrant closer CP surveillance, specialist referral, or disease-specific management, and which need a continued workup. 

"We hope our study will stimulate discussion about CP terminology, gene-disease associations, and variant interpretation," Arterbery says. "The most productive response will be further validation and development of better genotype-phenotype evidence rather than treating the findings as a definitive list of CP genes." 

Source: Arterbery AS, et al. (2026 Sep 3) Am J Hum Genet. A phenotypic paradigm for cerebral palsy genetics