Evaluation of the feasibility, diagnostic yield, and clinical utility of rapid genome sequencing in infantile epilepsy (Gene-STEPS): an international, multicentre, pilot cohort study

Background Most neonatal and infantile-onset epilepsies have presumed genetic aetiologies, and early genetic diagnoses have the potential to inform clinical management and improve outcomes. We therefore aimed to determine the feasibility, diagnostic yield, and clinical utility of rapid genome sequencing in this population. Methods We conducted an international, multicentre, cohort study (Gene-STEPS), which is a pilot study of the International Precision Child Health Partnership (IPCHiP). IPCHiP is a consortium of four paediatric centres with tertiary-level subspecialty services in Australia, Canada, the UK, and the USA. We recruited infants with new-onset epilepsy or complex febrile seizures from IPCHiP centres, who were younger than 12 months at seizure onset. We excluded infants with simple febrile seizures, acute provoked seizures, known acquired cause, or known genetic cause. Blood samples were collected from probands and available biological parents. Clinical data were collected from medical records, treating clinicians, and parents. Trio genome sequencing was done when both parents were available, and duo or singleton genome sequencing was done when one or neither parent was available. Site-specific protocols were used for DNA extraction and library preparation. Rapid genome sequencing and analysis was done at clinically accredited laboratories, and results were returned to families. We analysed summary statistics for cohort demographic and clinical characteristics and the timing, diagnostic yield, and clinical impact of rapid genome sequencing. Findings Between Sept 1, 2021, and Aug 31, 2022, we enrolled 100 infants with new-onset epilepsy, of whom 41 (41%) were girls and 59 (59%) were boys. Median age of seizure onset was 128 days (IQR 46–192). For 43 (43% [binomial distribution 95% CI 33–53]) of 100 infants, we identified genetic diagnoses, with a median time from seizure onset to rapid genome sequencing result of 37 days (IQR 25–59). Genetic diagnosis was associated with neonatal seizure onset versus infantile seizure onset (14 [74%] of 19 vs 29 [36%] of 81; p=0·0027), referral setting (12 [71%] of 17 for intensive care, 19 [44%] of 43 non-intensive care inpatient, and 12 [28%] of 40 outpatient; p=0·0178), and epilepsy syndrome (13 [87%] of 15 for self-limited epilepsies, 18 [35%] of 51 for developmental and epileptic encephalopathies, 12 [35%] of 34 for other syndromes; p=0·001). Rapid genome sequencing revealed genetic heterogeneity, with 34 unique genes or genomic regions implicated. Genetic diagnoses had immediate clinical utility, informing treatment (24 [56%] of 43), additional evaluation (28 [65%]), prognosis (37 [86%]), and recurrence risk counselling (all cases).


Introduction
Infantile-onset epilepsies range in severity from selflimited epilepsies to the larger group of developmental and epileptic encephalopathies. 1 The incidence of infantile-onset epilepsies is one in 1200. Patients with developmental and epileptic encephalopathies have drug-resistant seizures, severe developmental impairment, and high mortality risk, with important psychosocial implications for families and substantial economic costs for health systems. 1,2 Infantile-onset epilepsies often have genetic aetiologies (>800 genes implicated). 3 Numerous studies, including a systematic review, 4 show high diagnostic yield and costeffectiveness of gene panels and exome sequencing in early-onset epilepsies, with genetic testing now considered a first-line investigation. [5][6][7][8] Genome sequencing further increases diagnostic yield, 4 but has not been studied in unselected infantile epilepsy cohorts. In rare disease, genome sequencing, especially trio genome sequencing, has demonstrated substantial diagnostic yield. 9 For infants with epilepsy, the identification of a precise diagnosis can guide clinical management and inform prognosis regarding seizure control, developmental outcome, and potential comorbidities. A growing number of genetic epilepsies have precision treatment implications, including four common infantile epilepsy genes (KCNQ2, PRRT2, SCN1A, SLC2A1). 7 Although genetic therapies are not currently available for most epilepsies, tailoring of antiseizure medication is often possible. 10 Furthermore, genetic diagnoses could inform eligibility for clinical trials or non-antiseizure medication treatment (eg, epilepsy surgery) and enable precise genetic counselling. In a few studies of the effect of non-genome sequencing genetic testing in epilepsy, genetic diagnoses affected management in 36-72% of cases. [11][12][13][14][15] Although rapid genetic testing and prompt implementation of individualised treatment, where available, will possibly improve outcomes, a major challenge is that testing often takes months to years, with infants having progressive neurological sequelae from uncontrolled seizures or underlying disease. 16 Studies done in neonatal intensive care units (NICUs) and paediatric intensive care units (PICUs) demonstrate high diagnostic yield of rapid (ie, weeks) and ultrarapid (ie, days) genome sequencing for a range of conditions, with clinical utility and reduction in health-care costs. [17][18][19] To date, rapid genome sequencing has been undertaken primarily in ICUs, and the effect of prompt genetic diagnoses in infants with epilepsy has not been established. In this study, we therefore aimed to demonstrate the feasibility of rapid genome sequencing and investigate the diagnostic yield and clinical utility for infants with new-onset epilepsy.

Study design and cohort
We conducted an international, multicentre, cohort study (Gene-Shortening Time of Evaluation in Paediatric epilepsy Services [STEPS]), which is a pilot study of the International Precision Child Health Partnership (IPCHiP). This partnership is a consortium of

Research in context
Evidence before this study We searched PubMed using the terms "epilepsy" OR "seizure(s)" AND "rapid" AND "sequencing" for studies published from database inception to Jan 1, 2023, with no language restrictions. We identified case reports of rapid exome or genome sequencing in patients with epilepsy and several studies of rapid exome or genome sequencing in critically ill paediatric cohorts recruited from neonatal and paediatric intensive care units, including some participants with seizures. We also identified a recent systematic review of genetic testing in the epilepsies, which found the highest diagnostic yield for (non-rapid) genome sequencing (48%) followed by exome sequencing (24%). No studies of rapid exome or genome sequencing (ie, with results available within weeks) in epilepsy cohorts exist.

Added value of this study
We report an international, multicentre, cohort study of the feasibility, diagnostic yield, and clinical utility of rapid genome sequencing in 100 infants with new-onset epilepsy, using trio-based analyses when parental DNA was available. To date, this study is the first to evaluate rapid sequencing in a disease-specific cohort and the first study consisting of patients mostly outside an intensive care setting. First, we show that rapid genome sequencing has high diagnostic yield (43 [43%] of 100 infants) in infantile epilepsy and demonstrate the feasibility of rapid turnaround for participants recruited from intensive care, non-intensive care inpatient, and outpatient settings across multiple health-care systems. Second, we demonstrate marked genetic heterogeneity across our cohort and demonstrate the ability of rapid genome sequencing to identify genetic diagnoses missed by standardof-care genetic testing. Third, we observed that most parents of infants with newly diagnosed epilepsy are interested in rapid sequencing, and we demonstrate immediate clinical utility of genetic diagnoses for infants and their families in most cases.

Implications of all the available evidence
The findings from this study strongly support the implementation of rapid genome sequencing in the clinical evaluation of infants with new-onset epilepsy. These findings also enhance our understanding of underlying genetic mechanisms of epilepsy. Future research will be needed to understand the personal and long-term utility of early genetic diagnosis in infantile epilepsy. This study provides a framework for advancing precision health that can be implemented for other unexplained conditions beyond epilepsy.
four paediatric centres with tertiary-level subspecialty services, created to advance precision child health: Melbourne Children's Campus (MCC; Murdoch Children's Research Institute and The Royal Children's Hospital) in Australia; The Hospital for Sick Children (SickKids) in Canada; University College London Great Ormond Street Institute of Child Health (UCL GOS ICH) in the UK; and Boston Children's Hospital (BCH) in the USA.
We recruited infants with new-onset epilepsy or complex febrile seizures from the IPCHiP centres. Potentially eligible infants were identified by the study team and treating clinicians. The study team reviewed medical records and determined eligibility in discussion with treating clinicians. Infants younger than 12 months at seizure onset and recruited within 6 weeks of study site presentation were enrolled into the study with parental consent. We excluded infants with simple febrile seizures, acute provoked seizures, known acquired cause, or known genetic cause (ie, diagnostic genetic test result or clinical findings consistent with a monogenic syndrome, such as tuberous sclerosis complex). Brain MRI was reviewed to confirm lack of acquired aetiology at screening or as soon as available. We did not exclude infants with structural brain malformations without known genetic cause, or infants with a previous nondiagnostic or concurrent in-progress genetic testing, so as not to disrupt site-specific clinical standard of care. We worked with certified interpreters at each site for non-English-speaking families.
This study was approved by each site's institutional review boards and human ethics research committees. We obtained written informed consent from parents for research enrolment, clinically accredited rapid genome sequencing, and results return.

Clinical data
Clinical data were collected from medical records, treating clinicians, and parents. We documented study site, referral setting (outpatient, non-intensive care inpatient, NICU, PICU), sex, parent-reported race, gestational age, family medical history, epilepsy details (age at seizure onset, seizure type, EEG findings), develop ment before seizure onset, developmental plateau or regression following seizure onset, other neurological and non-neurological features, MRI findings, previous and concurrent genetic testing, and, if applicable, age at death. We classified epilepsy syndrome using the International League Against Epilepsy definitions, and we classified an epilepsy syndrome as other when the participant's presentation did not fit diagnostic criteria for one of those definitions. 1

Rapid genome sequencing
Blood samples were collected from probands and available biological parents. We did trio genome sequencing when both parents were available, and duo or singleton genome sequencing when one or neither parent was available. Site-specific protocols were used for DNA extraction, library preparation, genome sequencing, variant identification, and validation at clinically accredited laboratories (appendix pp 2-3). All sites performed genome-wide analysis for single nucleotide variants, small insertions and deletions, and copy number variants; the laboratory used by BCH was also clinically accredited to report short tandem repeat expansions in FMR1 and DMPK. Variant classification used standardised criteria (American College of Medical Genetics and Genomics 20 or Association for Clinical Genomic Science). Site-specific policies were followed for reporting variants of uncertain significance and secondary or incidental findings (appendix pp 2-3). Infants with pathogenic or likely pathogenic variants in genes consistent with phenotypes and modes of inheritance were considered to have diagnostic rapid genome sequencing. For infants with variants of uncertain significance that were plausibly explanatory (ie, no data ruled out pathogenicity, but insufficient data were present to classify as pathogenic or likely pathogenic variants), we reviewed medical records for clinical features or further investigations to support pathogenicity to deem variants clinically diagnostic.

Effect of rapid genome sequencing
We documented age at study site presentation, enrolment, blood collection, and rapid genome sequencing result. Short-term clinical utility (ie, to December, 2022) of rapid genome sequencing was assessed through medical records and treating clinicians. We defined clinical utility as actual influence on treatment, potential for precision therapy, additional investigation indicated or avoided, additional prognostic information, influence on goals of care, or influence on genetic counselling (beyond recurrence risk).

Statistical analysis
We analysed summary statistics for cohort demographic and clinical characteristics and the timing, diagnostic yield, and clinical effect of rapid genome sequencing. We analysed associations of demographic features, clinical features, and timing with diagnostic rapid genome sequencing using a two-tailed χ² test, Fisher's exact test, Mann-Whitney test, or Kruskal-Wallis test (based on normality assessment using Kolmogorov-Smirnov and Shapiro-Wilk tests) using the program SPSS (version 27.0), with statistical significance set at p<0·05.

Role of the funding source
The funders of the study had no role in study design, data collection, data analysis, data interpretation, writing of the report, or the decision to submit for publication.
Median time from seizure onset to site presentation was 7 days (IQR 1-24), from site presentation to enrolment was 3 days (1-9), from enrolment to proband sample collection was 0 days (0-1), and from sample collection to rapid genome sequencing result was 20 days (14-22; figure 2A). 91 (91%) of 100 families had trio genome sequencing, eight (8%) had duo genome sequencing, and one (1%) had singleton genome sequencing. Median study turnaround time from enrolment to rapid genome sequencing result was 21 days (IQR 15-23), shorter at one site (BCH) than the others (median 15 days vs 21-25 days; adjusted p<0·05 for pairwise comparisons) and not significantly different between referral settings. Median time from seizure onset to rapid genome sequencing result was 37 days (IQR 25-59) and median age at rapid genome sequencing result was 172 days (91-250), following median age at seizure onset of 128 days (appendix pp 4, 14-17).
In 15 cases, rapid genome sequencing identified genetic diagnoses not made by site-specific standard of care clinical testing (table 2): five (33%) with previous non-diagnostic testing and ten (67%) with concurrent non-diagnostic testing. In one infant, rapid genome sequencing detected a mosaic copy number variant (validated with karyotype) not identified on chromosomal microarray. In another infant, singleton gene panel identified a SCN2A variant classified as a variant of uncertain significance; trio rapid genome sequencing identified the variant as de novo, leading to the classification as likely pathogenic and facilitating immediate management changes.
Of the 57 infants with non-diagnostic genome sequencing, ten (18%) had variants of uncertain significance in genes potentially relevant to phenotypes (appendix pp [18][19][20][21][22][23][24]. Secondary or incidental diagnostic findings were detected in five (5%) of 100 infants (appendix pp [25][26]. Clinical utility was present for 42 (98%) of 43 infants with genetic diagnoses ( also affected. †Confirmed biochemically. ‡Considered clinically diagnostic by clinical team. §Subsequently found to have abnormal nerve conduction testing. ¶Parent from whom variant was inherited was suspected to also be affected. Data are n (%), unless otherwise specified. *All families received recurrence risk counselling based on the mode of inheritance of the diagnostic variants. Yes in this column refers to new health implication for parents or referral of additional family members for genetic testing for the diagnostic variants. †Implication for precision treatment based on the genetic aetiology regardless of whether the treatment was used. ‡Diagnosis did not have direct utility for this case as the infant died before the rapid genome sequencing result was available.

Discussion
To the best of our knowledge, this international, multicentre Gene-STEPS study is the first study of rapid genomic testing primarily outside an intensive care setting and in a disease-specific cohort. We demonstrate feasibility of rapid genome sequencing in infants with new-onset epilepsy across multiple tertiary paediatric systems, with high diagnostic yield and clinical effect. Our findings provide support to prompt the use of stateof-the-art rapid genomic testing to facilitate early aetiological diagnosis that can inform urgent targeted management in this vulnerable population. We demonstrate feasibility of expanding trio rapid genome sequencing from intensive care to outpatient and non-intensive care inpatient settings in four countries, with more than 80% of infants recruited from non-intensive care settings. More than 90% of parents consented, showing their interest in identifying the cause of their infant's seizures through early, rapid, and comprehensive genetic testing. Through the IPCHiP consortium, we harmonised study protocols across sites, strengthening the generalisability of our findings. Despite our sites having expertise in genomics and epilepsy, as well as institutional resources, this study posed challenges, including the cost of rapid genome sequencing and the need for sufficient personnel to efficiently achieve recruitment, research and clinical consent, sample collection, timely laboratory processes, variant interpretation, and return of results. Our experience highlights the need for collaboration between neurologists, geneticists, and genetic counsellors to ensure rapid identification of clinically significant variants to optimise patient care.
To our knowledge, this study is the first to evaluate rapid genome sequencing in infants with epilepsy. Our diagnostic yield of 43% is consistent with the yield of non-rapid genome sequencing (48%) in epilepsy reported in a recent systematic review (350 participants mostly with developmental and epileptic encephalopathies or severe phenotypes) and higher than that of chromosomal microarray (9%), gene panels (19%), and exome sequencing (24%), acknowledging that these studies have different inclusion or exclusion criteria. 4 We excluded infants with acquired epilepsies, who would be predicted to have far lower likelihood of genetic aetiologies, and infants with known genetic causes, whose inclusion would have increased the diagnostic yield of rapid genome sequencing. Overall, although our cohort is not population based, 21,22 it represents most infants who present to tertiary paediatric centres with unexplained epilepsy. Most of our findings are de novo and could thus be relevant to patients of all ancestries. Nonetheless, a limitation of our study is that most infants have parent-reported White race. Future studies including more diverse populations are needed to achieve broader generalisability.
We confirm high diagnostic yield in neonatal-onset epilepsies, self-limited epilepsies, and early infantile developmental and epileptic encephalopathies, with relatively lower, although still important, yield in infantile epileptic spasms syndrome. The varied yield for different epilepsy syndromes highlights the importance of rigorous phenotyping when counselling families. Indeed, in four (29%) of 14 infants with primary findings of variants classified as uncertain significance by standardised criteria, the variants of uncertain significance were considered clinically diagnostic by expert clinicians given phenotype-genotype correlation; in two (50%) of four cases, further clinical investigation confirmed pathogenicity.
We also confirm genetic heterogeneity and the importance of channelopathies (15% of cohort) in infantile-onset epilepsies. In contrast to previous studies utilising gene panels or exome sequencing, we did not see clear predominance of a small number of genes (eg, KCNQ2, PRRT2, and SCN1A). 4,7,22 This finding might reflect that previous studies using gene panels were limited to analysing specific subsets of genes or were conducted before the associations of other genes with epilepsy were identified. A potential limitation of our study is that our findings in a cohort of 100 participants might not reflect the full heterogenous genetic landscape of infantile-onset epilepsies. Furthermore, our study was not powered for a multivariate predictive model to assess which factors best predict a higher likelihood of identifying a genetic diagnosis; a larger cohort would be needed to develop such a model and investigate potential confounders in our analysis.
Genome sequencing represents the most comprehensive genetic testing approach but is not yet widely available. In most clinical settings, current standard of care includes chromosomal microarray or gene panel or exome sequencing (including tests performed on an exome sequencing or genome sequencing backbone using next-generation sequencing technology but analysed for only a small number of genes), performed concurrently or sequentially. Although our study was not designed to directly compare rapid genome sequencing with other tests, we demonstrate high yield of genome sequencing, performed as trio rapid genome sequencing whenever biological parents were available, and highlight its ability to detect genetic diagnoses not revealed by other modalities. Our findings support a genome-wide approach (exome sequencing or genome sequencing) as first-line genetic testing in infantile epilepsies, following guidelines endorsed by the American Epilepsy Society. 6 Future studies are needed to accurately quantify the additional yield of genome sequencing compared with other tests in epilepsy. 23 We anticipate that genome sequencing, which can detect single nucleotide variants, copy number variants, and other variant types, will become first-line testing and obviate the need for multiple tests in most patients, with trio rapid genome sequencing further enhancing yield. 24 Infants with epilepsy represent a vulnerable population with substantial morbidity and mortality burden. Unlike previous epilepsy cohort studies with exome sequencing or genome sequencing performed in research laboratories, [25][26][27] we performed rapid genome sequencing in clinically accredited laboratories, allowing immediate return of results to families and clinicians. Clinical utility was present for 55% of the cohort, including 98% with diagnostic rapid genome sequencing and 23% with non-diagnostic rapid genome sequencing or secondary or incidental findings. For participants with diagnostic rapid genome sequencing, we report a higher rate of clinical utility than with previous studies. [11][12][13][14][15] Because of the current follow-up duration, we can only report short-term utility; additional utility is likely to be observed long term. We encourage future studies to report utility of non-diagnostic and secondary or incidental findings, as we found meaningful utility in multiple cases.
We identified numerous positive effects of early genetic diagnosis, affecting treatment (56%), evaluation (65%), and prognostic counselling (86%), and suggesting potential precision therapies (49%). In some cases, genetic diagnosis suggested a relatively good prognosis, with high likelihood of weaning antiseizure medication and normal development (eg, PRRT2). In other cases, genetic diagnosis suggested a relatively poor prognosis, with high likelihood of drug-resistant seizures, global developmental delay or intellectual disability, and even early mortality (eg, BRAT1), thus informing goals of care. Making a precise diagnosis also guides recurrence risk counselling, whether with inherited variants (high risk) or with apparently de novo variants (low but not zero risk due to the inability to detect parental gonadal mosaicism 28 ), which is important to guide reproductive decision making for families.
We acknowledge some negative or difficult aspects of rapid genome sequencing. Early genetic diagnosis and awareness of future prognosis might contribute to diagnostic shock and parental stress. 29,30 As rapid genome sequencing becomes more widespread, families should be counselled before consenting and supported after results are returned. An important issue is the variable severity of conditions associated with a single gene. For example, KCNQ2, KCNQ3, SCN1A, SCN2A, and SCN8A are associated with phenotypes ranging from self-limited epilepsies with normal developmental outcome to intermediate severity conditions to drug-resistant epilepsies with profound developmental impairment. 2 Precise prognostication is not always possible early on, and this uncertainty is very challenging for families. Our findings also included several neurodevelopmental disorders in which develop mental impairments and other clinical features might become evident after infancy. Longitudinal evaluation is essential to monitor for additional clinical features and delineate the genetic landscape of epilepsy with and without neuro developmental disorders. 26,31 An additional area of uncertainty relates to the detection of variants of uncertain significance not considered clinically diagnostic, as was the case for 10% of our cohort, which might require time or additional investigation to resolve. Detection of variants of uncertain significance is a feature of all genomic tests given that our knowledge of the genome and disease associations is incomplete. Exome sequencing or genome sequencing, especially trio, might be associated with fewer variants of uncertain significance than gene panel testing. 32 Finally, trio rapid genome sequencing can identify secondary or incidental diagnostic findings in the infant and, thus, parents, as occurred in 5% of our cohort. Parents require adequate pre-test counselling regarding this possibility and post-test support for coping with unexpected familial health implications.
The turnaround time of rapid genome sequencing has recently been reported to be on the order of hours in intensive care settings, compared with weeks in our study, suggesting room for improvement. 33 However, given the inclusion of participants from both inpatient and outpatient settings, and the baseline lack of access to rapid-or any-genomic sequencing for many participants, a median turnaround time of 21 days from study enrolment to rapid genome sequencing result represents a major improvement over current standard of care. Moreover, although we aimed to perform trio genome sequencing for all individuals, for nine (9%) of 100 infants we were only able to perform duo or singleton genome sequencing. Although this approach might reduce opportunities for genetic diagnosis and discovery relative to trio testing, we believe these options are essential for ensuring equitable access when biological parents are unavailable.
We focused on initial diagnostic yield and short-term impact of rapid genome sequencing. Longitudinal follow-up will be essential to demonstrating the importance of rapid diagnosis in improving clinical, quality of life, and economic outcomes, which will inform advocacy and policy decisions about funding of genetic testing. Further aspects to assess include the parental perspective regarding rapid genome sequencing to ensure acceptability for those most likely to benefit from early diagnoses, reanalysis of genome sequencing data to increase diagnostic yield over time, and implementation of rapid genome sequencing into routine clinical practice.
We demonstrate the success and effect of a collaborative international model to provide rapid genetic diagnosis and clinical utility to infants with epilepsy through prospective enrolment, phenotyping, rapid genome sequencing, interpretation, and return of results. The diagnostic yield and short-term clinical effects are already high, and we anticipate long-term benefits for patients and families. As we shift the paradigm of epilepsy evaluation and diagnosis in the first year of life, this model might serve as a blueprint for advancing precision health for additional diseases whose aetiologies are suspected to be genetic but remain largely unexplained.