LEVETIRACETAM VS. PHENOBARBITAL FOR NEONATAL SEIZURES: A RETROSPECTIVE COHORT STUDY

1 Objectives: Although phenobarbital (PB) is commonly used as a first-line anti-seizure 2 medication (ASM) for neonatal seizures, in 2015 we chose to replace it with levetiracetam 3 (LEV), a third-generation ASM. Here, we compared the safety and efficacy of LEV and PB 4 as first-line ASM, considering the years before and after modifying our treatment protocol. 5 Methods: We conducted a retrospective cohort study of 108 neonates with EEG-confirmed 6 seizures treated with first-line LEV or PB in 2012-2020. 7 Results: First-line ASM was LEV in 33 (31%) and PB in 75 (69%) neonates. The etiology 8 included acute symptomatic seizures in 69% of cases (30% hypoxic-ischemic 9 encephalopathy, 32% structural vascular, 6% infectious, otherwise metabolic), and neonatal 10 epilepsy in 22% (5% structural due to brain malformation,17% genetic). Forty-two of 108 11 (39%) neonates reached seizure freedom following first-line therapy. Treatment response 12 did not vary by first-line ASM among all neonates, those with acute symptomatic seizures, or 13 those with neonatal onset epilepsy. Treatment response was lowest for neonates with a 14 higher seizure frequency, particularly for those with status epilepticus vs. rare seizures 15 ( p <0.001), irrespective of gestational age, etiology, or EEG findings. Adverse events were 16 noted in 22 neonates treated with PB and in only one treated with LEV ( p <0.001). 17 Conclusion: Our study suggests a potential non-inferiority and a more acceptable safety 18 profile for LEV that may thus be a reasonable option as first-line ASM for neonatal seizures 19 in place of PB. Treatment should be initiated as early as possible since higher seizure 20 frequencies predispose to less favorable responses. 25 26 27 28 29 30

We retrospectively identified from our institutional database at the University Children's 10 Hospital Zurich preterm and term neonates (≤30 days of age, corrected gestational age ≤44 11 weeks) with seizures diagnosed and treated from December 2011 to July 2020. Only 12 neonates 1) with EEG-confirmed seizures, 2) with either LEV or PB as first-line ASM, and 3) 13 with available informed general consent were considered for this study. We excluded 14 neonates 1) who received first-line treatment with an ASM other than LEV or PB, and 2) 15 whose parents had not given informed general consent. The sequence of ASM

24
Neonates underwent scalp video-EEG with 12 or 21 electrodes placed according to the 25 international 10-20 system, depending on the infant's head circumference. Extracerebral 26 leads were applied for respiratory and electrocardiogram recording and surface asynchrony for age, and 3) severely abnormal = isoelectric or low voltage invariant activity, 11 burst-suppression pattern, permanent discontinuous activity.

13
Electrographic seizures in neonates were defined as sudden, abnormal EEG events with a 14 repetitive and evolving pattern with a peak-to-peak voltage of >2μV and a duration of >10sec

15
[32], while "evolving" refers to an unequivocal evolution in frequency, voltage, morphology, or location. Seizures were classified as electrographic only if no clinical signs were observed by the bedside provider or on video review, or electroclinical if a clinical sign was associated with an ictal discharge [33]. Seizure etiology was classified based on the current framework for neonatal seizures and epilepsy syndromes [33] as hypoxic-ischemic, structural vascular 20 (Including acute ischemic stroke, hemorrhage and other vascular induced ischemia), 21 structural due to brain malformation, genetic, infectious, metabolic, and unknown. We further divided seizure etiology to acute symptomatic seizures, including hypoxic-ischemic, 23 structural vascular, infectious, and metabolic, and neonatal epilepsies, including structural 24 due to brain malformation and genetic.

25
Seizure frequency within the first 24 hours was determined by clinical reports, and findings of 26 continuous monitoring by EEG/aEEG if available, as 1) rare (<5 seizures), 2) occasional (5-  doses received maintenance PB at 2.5 mg/kg per dose, given IV once daily.

16
Treatment response was defined as 1) complete cessation of EEG seizures following ASM 17 administration, and 2) no administration of further ASM for seizure control. If ≥2 ASM were 18 administered before seizure cessation, the last administered ASM was defined as effective.

19
Adverse events were noted at the presence of hypotension, heart rate or respiratory 20 abnormalities with need for oxygen, ventilation, or vasopressor treatment, irritability or 21 sedation with poor feeding, and changes in laboratory parameters (complete blood cell 22 count, electrolytes, liver enzymes, ammonia, and blood gas analysis) that were attributed to 23 an ASM by the clinical team and documented in the medical record.  according to dosage were performed with Fisher's exact test for categorical variables (two-3 sided, no correction for multiple testing) and Wilcoxon rank-sum test for continuous 4 variables. We fitted a quasi-Poisson generalized linear model with log link to assess the 5 incidence risk ratio of incomplete treatment response to the first-line and second-line ASM, 6 respectively. We included gestational age, etiology, seizure frequency, EEG findings, and 7 ASM as explanatory variables. We excluded 9 neonates with seizures of unknown etiology 8 from this analysis. Statistical analysis was performed with R version 4.1.2 (R Foundation for 9 Statistical Computing, Vienna, Austria). We did not account for missing data in our analysis.

Treatment response according to etiology
3 Among all neonates with seizures (n=108), only 39% reached seizure freedom following 4 first-line therapy: treatment response did not vary by ASM (p = 0.40, Figure 1).

5
Among neonates with acute symptomatic seizures (n=75), only 39% reached seizure 6 freedom following first-line therapy: treatment response did not vary by ASM (p = 0.59,

24
Adverse events were noted in 22 (24%) of neonates treated with PB and in only 1 (1%) of overall low in our study and did not differ between LEV and PB among all neonates, or 3 among those with acute symptomatic seizures and neonatal onset epilepsy. Treatment 4 response was lowest for neonates with a higher seizure frequency but did not differ by gestational age, etiology, or EEG findings. The adverse-effect profile of LEV was more 6 favorable compared to PB. Our study thus suggests that LEV may be a safe and effective 7 alternative to PB as a first-line ASM in treating neonatal seizures.  study design may at least partly account for the divergent results between our study (and 10 other past studies) and the recent trial that reported a considerably higher efficacy for PB 11 than for LEV [28].

12
Neonates with a higher seizure frequency in our study were less likely to respond to the first-  one-fourth each of our cohort, may at least partly account for the relatively low rates of 21 seizure cessation after first-line ASM administration in our study, compared to previous work 22 [18,28]. Interestingly, higher ASM doses were not linked to improved efficacy in our cohort, 23 although some neonates received LEV dosages higher than the licensed 60mg/kg per day 24 that have recently been shown to improve seizure control [43]. Although it is tempting to try 25 higher doses, particularly of LEV, due to its favorable safety profile, to overcome incomplete seizure response, this should not lead to unnecessary delays in treatment with other, 1 potentially effective ASM.

ASM adverse events
3 Adverse events, including hypotension, respiratory suppression, and sedation, were noted in 4 one-fourth of neonates treated with PB and only in a single case of those treated with LEV.

5
Our study thus supports a more favorable adverse-effect profile of LEV than PB, in line with 6 previous work [18,28]. In particular, hypotension attributed to PB administration has been