Genetic testing and hospital length of stay in neonates with epilepsy

Abstract

Objective:

We evaluated changes in genetic testing for neonatal onset epilepsy and associated short-term outcomes over an 8-year period among a cohort of NICU patients at a single institution before and after the introduction of sponsored genetic epilepsy testing in January, 2018.

Methods:

Our primary outcome was a change in length of stay (LOS) after 2018. We also ascertained severity of illness with the Neonatal Sequential Organ Failure Assessment (nSOFA), type and result of genetic testing, turnaround time to molecular diagnosis (TAT), LOS, antiseizure medications (ASM) and use of technology at discharge. We compared outcomes using non-parametric tests and difference-in-difference analysis.

Results:

Fifty-three infants with genetic testing were included; 20 infants were tested after 2018. 4,160 infants in the NICU without genetic testing were used as reference. In the genetic testing group, LOS was 25 days (IQR 5, 49) pre-2018 and 19 days (IQR 6, 19) post-2018 (p<0.001 when compared to the reference population in the difference-in-difference analysis). TAT decreased from 51 days to 17 days after 2018 (p=0.003). ASM number decreased from 4 (IQR 2,5) to 2 post-2018 (IQR1,3) (p=0.02). Over the same time periods there was no significant change in birth weight, maximum nSOFA score, or technology dependence.

Conclusions:

In this cohort, changes in genetic testing for neonatal onset epilepsy were associated with shorter LOS that was not explained by changes in severity of illness, birth weight, or the average LOS in the NICU over time. Validation of these results in a larger, multicenter sample size is warranted.

Introduction

Neonatal onset seizures are one of the most frequent symptoms evaluated in the neonatal intensive care unit (NICU). Incident rates are up to 2.6 cases per 1000 live births ^1–3. Most neonatal onset seizures are acute symptomatic, secondary to conditions such as metabolic or electrolyte disorders, hypoxic ischemic injury, intracerebral hemorrhage, perinatal stroke, or intracranial infections ^4,5. Approximately 15–20% of neonates have neonatal epilepsy much of which is now recognized to be secondary to monogenic conditions. Genetic epilepsies are heterogenous and lead to a wide range of clinical diseases. Organic acidurias, peroxisomal disorders, urea cycle defects, vitamin deficiencies, ion channelopathies, and malformations of cortical development are some of the genetic conditions associated with neonatal seizures ^6–8. The rapid identification of a monogenic condition may impact treatment and may provide information for families regarding recurrence risk ⁹.

There are no guidelines for a standardized approach for genetic testing of neonatal onset epilepsies. Prior studies for early onset epilepsies have demonstrated an increasing yield for genetic testing and its cost effectiveness ^10,11. The decision of whether to send genetic testing for the evaluation of neonatal onset epilepsies and the type of test sent is provider dependent. Over the past decade, the modalities of genetic testing available have evolved and include microarrays, single gene tests, next generation sequencing panels, whole exome sequencing and in some hospitals, whole genome sequencing. Importantly, the cost of genetic testing has fallen substantially over the past few years potentially impacting the role in the diagnosis of neonatal epilepsies ^12,13. In addition to greater availability and lower cost, there has also been rapid expansion in our knowledge of the number of genes that are causally associated with epilepsy ^9,14–16.

Despite the emergence of new modalities of available genetic testing and an expansion of knowledge of genetic conditions associated with neonatal epilepsy, little is known about the evolution of genetic testing patterns in the NICU nor the impact of genetic testing and the ability to establish a molecular diagnosis on short-term outcomes for neonates with epilepsy. The aim of this study was to analyze genetic testing patterns over the period of 2012 through 2020 in neonates with seizures admitted to a single center NICU and to determine whether genetic testing affects short-term outcomes.

Methods

We performed a retrospective observational cohort study of neonates presenting with new onset epilepsy who received genetic testing while being treated in the neonatal intensive care unit (NICU) at the Ann & Robert H. Lurie Children’s Hospital of Chicago between June, 2012 and October, 2020. Neonates with ICD-9 and ICD-10 codes for neonatal seizures and confirmed electrographic seizures who received genetic testing (microarray, karyotype, single gene testing, next generation sequence testing [NGS], or whole exome sequencing [WES]) were included in the analysis. Genetic testing was ordered by the treating health care providers (neonatologists, neurologists, geneticists) based on clinical indications. Patients with neonatal onset seizures determined to be acute symptomatic seizures or infants for whom genetic testing was performed after discharge from the NICU were excluded. The primary outcome was length of stay (LOS) and turnaround time for receiving genetic test results for a cohort that underwent genetic testing prior to December 31^st, 2017 and compared it to a cohort that had testing after January 1^st, 2018 when sponsored NGS genetic testing for epilepsy became routinely available, in addition to other clinically available tests. As a reference group for the LOS analysis, we included all neonates admitted to the NICU during the same time period who had no genetic testing for the evaluation of seizures. Secondary outcomes included number of genetic tests performed, number of antiseizure medications (ASM) used during the hospital stay, number of ASMs at discharge, and the use of tracheostomy and feeding tubes at discharge. The local Institutional Review Board on Human Research approved a waiver of consent for data collection (IRB 2020–3823).

Study data were collected from the electronic health record (EHR) and managed using REDCap electronic data capture tools hosted at Northwestern University ^17,18. Data collected included sex, race, gestational age, birth weight, Apgar scores, socioeconomic status, imaging and electroencephalographic data, type and number of ASMs, and tracheostomy and tube feeding at discharge, as well as type, dates and results of genetic testing. In order to account for severity of illness or changes in management across time periods, illness severity was ascertained using the neonatal Sequential Organ Failure Assessment (nSOFA) score on admission day one and maximum illness severity scoring, as previously published ¹⁹. The nSOFA score ranges from 0 to 15 and is calculated using the presence of mechanical ventilation and peripheral oxygen saturation/fraction of inspired oxygen ratio for respiratory dysfunction, presence of vasoactive medications and/or corticosteroids for cardiovascular dysfunction, and platelet count for hematologic function. This score has been validated across different NICUs and has shown to have good-to-excellent discrimination for all-cause mortality in neonates ²⁰.

Demographic data were summarized as medians with interquartile range (IQR). Descriptive comparisons were done using Fisher’s exact test and non-parametric Mann-Whitney tests using GraphPad Prism (version 9.1.2 for Mac, GraphPad Software, San Diego, California USA). Correlation between TAT and LOS was assessed with Pearson’s R. Difference-in-difference analysis based on interaction between type of genetic testing group and time period was performed using linear regression for birth weight, negative binomial regression for nSOFA score, and Poisson regression for length of stay (LOS) based on the distribution of the variables using R version 3.6.2 (R Foundation for Statistical Computing). Statistical significance was defined as a p-value of <0.05.

Results

A total of 53 neonates (49% male, n=26) with seizures and genetic testing were included in the primary analysis. Of those, 33 neonates were admitted before 2018. Demographic and clinical characteristics of both groups are summarized in Table 1.

Table 1.

Demographic Characteristics for Infants With Neonatal-Onset Epilepsy Who Received Genetic Testing

	2012–2017 (n=33)	2018–2020 (n=20)	p-value
Sex	45% male (n=15)	55% male (n=11)	0.58
Race
White	42% (n=14)	30% (n=6)
African American	21% (n=7)	35% (n=7)
Asian	6% (n=2)	10% (n=2)
Hispanic/Other	30% (n=10)	25% (n=5)
GA weeks, median, (IQR)	39 (37.7, 40)	39 (38,39)	0.92
Apgar Score 1 minute, median, (IQR)	8 (4,9)	8 (3.5,9)	0.80
Apgar Score 5 minutes, median, (IQR)	9 (6,9)	9 (7,9)	0.82
nSOFA day 1, median, (IQR)	0 (0,0)	0 (0,0)	0.80
ASM during NICU stay, median (IQR)	4 (2,5)	2 (1,3)	0.02
ASM at discharge, median (IQR)	2 (1,3)	1 (1,2)	0.10
Number of genetic tests, median, (IQR)	2 (1,2)	1 (1,2)	0.08
Type of testing n, (%)
Microarray	18 (54)	7 (35)	0.26
Karyotype	4 (12)	1 (5)	0.64
Single Gene Testing	9 (27)	3 (15)	0.50
NGS testing	23 (70)	15 (75)	0.76
WES	5 (15)	2 (10)	0.70
Tracheostomy at discharge n, (%)	2 (6)	2 (10)	0.63
Tube feedings at discharge n, (%)	19 (58)	10 (50)	0.78

Most infants with neonatal onset epilepsy were born at 39 weeks (IQR 38,40), had median 1-minute Apgar score of 8 (IQR 4, 9) and 5-minute Apgar score of 9, (6.7, 9), a day 1 nSOFA score of 0 (IQR 0,0) and a max nSOFA score of 0 (IQR 0,2). This did not vary across the two time-periods studied. Patients (n=4160) admitted in the NICU during the same time period who had no genetic testing for seizures were used as a reference population for the LOS analysis.

The utilization of different genetic tests did not vary substantially from before to after 2018: microarray: 15 (47%) versus 7 (35%), p=0.26; karyotyping: 3 (9%) versus 4 (20%), p=0.064; single gene testing 7 (22%) versus 3 (15%), p=0.50; NGS panel: 23 (71%) versus 15 (75%), p=0.76; and WES: 4 (13%) versus 2 (10%), p=0.70. Before 2018, however, most neonates received a median of 2 genetic tests whereas in 2018 and beyond, most infants received a single genetic test (p=0.08). Specifically, before 2018, 36% (n=12) received a single test and 64% (n=21) received 2 or more tests. After 2018, 60% (n=12) received a single test and 40% (n=8) received 2 or more tests. Molecular diagnostic yield was also similar across the two time periods. Overall, a pathogenic variant was identified in 11% (n=6) of microarrays, 2% (n=1) of karyotype analyses, 15% (n=8) of single gene tests, 23% (n=12) of NGS panels, and 9% (n=5) of WES.

The distribution of testing per year and number of patients evaluated is summarized in Supplementary Figure 1. The time to reach a molecular diagnosis regardless of genetic testing modality decreased from a median 51 days (IQR 24,80), to 17 days (IQR 10,24) in 2018 and beyond (p=0.003), although the turnaround time for genetic testing appeared to decline substantially over the study time period, it did not correlate with the LOS (Pearson’s R=0.0924, p=0.6272) (Figure 1).

Figure 1.

Time to molecular (pathogenic diagnosis) (n=29), by type of testing performed in the cohort; open symbols represent patients in whom a single genetic testing was used, solid symbols represent patients in whom more than 2 genetic tests were used to determine a pathogenic finding (R² 0.3993 p=0.0002)

Of twenty-nine patients (55%) for whom a pathogenic variant was found, the most common pathogenic change involved SCN2A (n=6), KCNQ2 (n=3), and GLDC (n=3) (Supplemental table 1).

Most infants were treated with a median of 3 ASMs (IQR 1.5, 4.5), and the most commonly used ASM was phenobarbital in 73% (n=38). Prior to 2018, neonates received a median of 4 ASMs (IQR 2, 5) during their NICU stay compared to 2018 and beyond when they received a median of 2 ASMs (IQR 1,3; p=0.02). We further considered the number of ASMs received in patients that were diagnosed with a sodium or potassium channelopathy (n=9). Prior to 2018 patients with a channelopathy received a median of 4.5 ASMs in comparison to after 2018, where they received a median of 3 ASMs (p=0.21).

We sought to ascertain differences associated with the change in genetic testing availability in length of stay (LOS), time to molecular diagnosis, number of antiseizure medications at discharge, and use of tracheostomy or tube feedings at discharge. Prior to 2018, the median LOS was 25 days (IQR 5, 49) and after 2018 was 19 days (IQR 6, 19) for infants that received genetic testing for the work up of neonatal onset epilepsy. To assess whether these differences in LOS could be due to changes in clinical practice, patient characteristics, or illness severity across the two time periods, we evaluated LOS, birthweight and maximum nSOFA score in the whole NICU population during both time periods with a difference-in-difference analysis. When compared to the entirety of the NICU population during the same periods of time, the median LOS was 9 days (IQR 3,35) prior to 2018, and 11 days (IQR 3, 38) after 2018. This difference-in-difference analysis using the rest of the NICU population as reference demonstrated a statistically significant change in LOS when genetic testing was performed for the evaluation of seizures after 2018 (p<0.001). During that same time period, the difference-in-difference in birth weight and maximum severity of illness (maximum nSOFA score) was not statistically significant, suggesting that changes in birth weight or severity of illness in the cohorts did not account for the observed changes in LOS in the genetic testing groups over time (Table 2).

Table 2.

Differences-in-Differences Analysis for Infants With Neonatal-Onset Epilepsy With Genetic Testing and Comparison of a NICU Reference Population Without Genetic Testing

Variable	2012–2017		2018–2020		Difference-in-differences p-value*
	Genetic testing		Genetic testing
	No (n=2893)	Yes (n=33)	No (n=1267)	Yes (n=20)
Birth weight in g, mean (SD)	2569 (1048)	3182 (874)	2561 (1061)	3242 (665)	0.82
Max. nSOFA score, median (IQR)	0 (0, 3)	0 (0, 2)	0 (0, 3)	0 (0, 2)	0.47
LOS in days, median (IQR)	9 (3, 35)	25 (15, 49)	11 (3, 38)	19 (6, 19)	<0.001

^*Differences-in-differences based on interaction between genetic testing group and time period using linear regression for birth weight, negative binomial regression for nSOFA score, and Poisson regression for length of stay (LOS) based on the distribution of the variables.

Discussion

Neonatal seizures are associated with an increased risk of death and severe long-term morbidity, including epilepsy, intellectual disability, and other neurological conditions ²¹. A precision medicine approach has become increasingly available thanks to the increased use of NGS targeted epilepsy gene panels and WES for genes that have diagnostic and therapeutic implications as is the case for sodium channelopathies, and may play an important role in counseling for disease prognosis and family planning counseling ^22–24.

The availability of sponsored genetic testing has lifted significant financial barriers making genetic testing more accessible. However, the short-term benefits of genetic testing have not been ascertained. Our study describes the use of genetic testing in a large academic institution and evaluated the impact of sponsored testing after 2018. Our data show that the availability and use of genetic testing, particularly of sponsored NGS panels, was associated with decreased LOS after adjustment for severity of illness during NICU admission and birth weight across an 8-year period. These results may indicate that the broad availability of genetic testing allowing for a more rapid and precise diagnosis, reducing the turnaround time for results and may inform a more precise therapeutic intervention, avoiding the need to perform multiple tests while waiting for results. Our results showed a trend towards a reduction in number of tests to obtain a molecular diagnosis and a significantly shorter turnaround time. Although it may be possible that the observed decrease in TAT before and after 2018 may have contributed to a decrease in LOS, the weak correlation between TAT and LOS in this cohort makes this explanation unlikely; suggesting an additional benefit for the use of genetic testing beyond shortening TAT. Interestingly, we observed a trend towards the use of fewer ASMs, possibly because the molecular diagnosis informed treatment choice in cases of sodium or potassium channelopathies as one of the most common findings in this cohort and consistent with previously reported studies⁶. There may, however, be other factors not accounted for in this study, such as therapeutic trends for the management of neonatal seizures, a topic outside of the scope of this study.

Our study has strengths and limitations. Among the limitations of the study, we have a small number of subjects in a single institution which preclude a robust regression analysis, as well as the retrospective data acquisition. In terms of strengths, as in previous studies ^6,23, we too found a preponderance of molecular diagnoses involving the KCNQ2 and SCN2A genes. Such findings have specific therapeutic implications for choice of ASM for the treatment of neonatal onset epilepsies. In consensus with currently available literature, our data is consistent with a high diagnostic yield ²⁵. Overall, these results support the evidence that currently available genetic testing can guide therapy in patients with neonatal onset epilepsies. It is possible, that the use of less widely available modalities of genetic testing such as whole genome sequence may further change short term outcomes as they become more clinically available and feasible ^26,27. Our study demonstrates an impact on length of stay with the increased use of routinely available genetic testing for the evaluation of neonatal onset epilepsies. Although our study is retrospective in nature, it reflects the population of a large urban center level IV NICU with a large number of neonates with epilepsy. Notably, a large referral center may not entirely represent the population of other less acute units where infants with neonatal onset seizures are admitted. Nonetheless, using a validated severity of illness score such as the nSOFA, may help to objectively ascertain the severity of illness across a number of years and across other centers as prospective multicenter studies are needed to confirm our findings.

Supplementary Material

1

Supplementary Figure 1.

Legend: Distribution of type of testing per year and number of patients tested.

Supplementary table 1.

Abbreviations:

NICU

Neonatal Intensive Care Unit

ASM

Antiseizure medications

LOS

Length of Stay

nSOFA

Neonatal Sequential Organ Failure Assessment

NGS

Next Generation Sequence

SGT

Single Gene Testing

WES

Whole Exome Sequence