Fenfluramine increases survival and reduces markers of neurodegeneration in a mouse model of Dravet syndrome

Abstract Objective In patients with Dravet syndrome (DS), fenfluramine reduced convulsive seizure frequency and provided clinical benefit in nonseizure endpoints (e.g., executive function, survival). In zebrafish mutant scn1 DS models, chronic fenfluramine treatment preserved neuronal cytoarchitecture prior to seizure onset and prevented gliosis; here, we extend these findings to a mammalian model of DS (Scn1a +/− mice) by evaluating the effects of fenfluramine on neuroinflammation (degenerated myelin, activated microglia) and survival. Methods Scn1a +/− DS mice were treated subcutaneously once daily with fenfluramine (15 mg/kg) or vehicle from postnatal day (PND) 7 until 35‐37. Sagittal brain sections were processed for immunohistochemistry using antibodies to degraded myelin basic protein (D‐MBP) for degenerated myelin, or CD11b for activated (inflammatory) microglia; sections were scored semi‐quantitatively. Apoptotic nuclei were quantified by TUNEL assay. Statistical significance was evaluated by 1‐way ANOVA with post‐hoc Dunnett's test (D‐MBP, CD11b, and TUNEL) or Logrank Mantel‐Cox (survival). Results Quantitation of D‐MBP immunostaining per 0.1 mm2 unit area of the parietal cortex and hippocampus CA3 yielded significantly higher spheroidal and punctate myelin debris counts in vehicle‐treated DS mice than in wild‐type mice. Fenfluramine treatment in DS mice significantly reduced these counts. Activated CD11b + microglia were more abundant in DS mouse corpus callosum and hippocampus than in wild‐type controls. Fenfluramine treatment of DS mice resulted in significantly fewer activated CD11b + microglia than vehicle‐treated DS mice in these brain regions. TUNEL staining in corpus callosum was increased in DS mice relative to wild‐type controls. Fenfluramine treatment in DS mice lowered TUNEL staining relative to vehicle‐treated DS mice. By PND 35‐37, 55% of control DS mice had died, compared with 24% of DS mice receiving fenfluramine treatment (P = 0.0291). Significance This is the first report of anti‐neuroinflammation and pro‐survival after fenfluramine treatment in a mammalian DS model. These results corroborate prior data in humans and animal models and suggest important pharmacological activities for fenfluramine beyond seizure reduction. Plain Language Summary Dravet syndrome is a severe epilepsy disorder that impairs learning and causes premature death. Clinical studies in patients with Dravet syndrome show that fenfluramine reduces convulsive seizures. Additional studies suggest that fenfluramine may have benefits beyond seizures, including promoting survival and improving control over emotions and behavior. Our study is the first to use a Dravet mouse model to investigate nonseizure outcomes of fenfluramine. Results showed that fenfluramine treatment of Dravet mice reduced neuroinflammation significantly more than saline treatment. Fenfluramine‐treated Dravet mice also lived longer than saline‐treated mice. These results support clinical observations that fenfluramine may have benefits beyond seizures.


| INTRODUCTION
Dravet syndrome (DS) is a rare, severe, treatment-resistant developmental and epileptic encephalopathy. 1,2More than 85% of patients with DS present with pathogenic variants in the SCN1A gene, 2 which encodes the alpha-1 subunit of the Na v 1.1 neuronal voltage-gated sodium channel. 3,4a v 1.1 is enriched in inhibitory gamma-aminobutyric acid (GABA)-ergic interneurons of the hippocampus and cortex. 5In DS, Na v 1.1 haploinsufficiency impairs inhibitory GABAergic action potentials in forebrain structures, and results in deficits in structure and function in the hippocampal network. 5,6In addition to causing seizures, Na v 1.1 haploinsufficiency causes nonseizure comorbidities as a direct consequence of the underlying pathology rather than-or in addition to-being caused only by neurological damage secondary to seizures. 7n patients with DS, the antiseizure medication fenfluramine has proven, clinical benefit in reducing convulsive seizure frequency [8][9][10][11] and alleviating nonseizure comorbidities, including survival 12 and regulation of behavior, emotions, and cognition (i.e., executive function). 8,13Additional preclinical evidence in nonseizure rodent models supports a role for fenfluramine in improving spatial learning and memory-important aspects of cognition. 14,15In the scn1 mutant zebrafish model of DS, aberrant neuronal cytostructural architecture with loss of GABAergic dendritic arborization occurred prior to seizure onset; in scn1 mutants, fenfluramine treatment restored GABAergic dendritic arborization and gliosis to normal, wild-type levels. 16Further, in the DBA/1 audiogenic seizure mouse model of sudden unexpected death in epilepsy (SUDEP), fenfluramine treatment reduced seizure-induced respiratory arrest (S-IRA) prior to SUDEP at a dose that did not significantly change seizure frequency compared to controls. 17Taken together, these results suggest that fenfluramine has additional, potentially disease-modifying activities beyond seizure control.
The complex pharmacology of fenfluramine is likely responsible for affecting these seizure and nonseizure morbidities.9][20] In vivo evidence supports agonist activity at the 5-HT 4 receptor 21 -a receptor associated with learning and memory in clinical studies. 22gnificance: This is the first report of anti-neuroinflammation and pro-survival after fenfluramine treatment in a mammalian DS model.These results corroborate prior data in humans and animal models and suggest important pharmacological activities for fenfluramine beyond seizure reduction.
Plain Language Summary: Dravet syndrome is a severe epilepsy disorder that impairs learning and causes premature death.Clinical studies in patients with Dravet syndrome show that fenfluramine reduces convulsive seizures.
Additional studies suggest that fenfluramine may have benefits beyond seizures, including promoting survival and improving control over emotions and behavior.Our study is the first to use a Dravet mouse model to investigate nonseizure outcomes of fenfluramine.Results showed that fenfluramine treatment of Dravet mice reduced neuroinflammation significantly more than saline treatment.
Fenfluramine-treated Dravet mice also lived longer than saline-treated mice.
These results support clinical observations that fenfluramine may have benefits beyond seizures.

K E Y W O R D S
Dravet syndrome, fenfluramine, myelin, neuroinflammation, survival

Key Points
In an Scn1a +/− mouse model of Dravet syndrome, fenfluramine treatment: • Protected against myelin damage in the hippocampus CA3 and parietal cortex.
Additionally, fenfluramine positively modulates the activity of the Sigma1 receptor (Sigma1R), potentially in conjunction with endogenous neuro(active) steroids. 14,15ouse models of DS recapitulate features of the DS clinical phenotype, including nonseizure comorbidities such as alterations in emotions (e.g., anxiety) and deficits in behavior, cognition, and social interaction, 23 as well as spontaneous seizures and increased mortality. 24hese phenotypic changes are accompanied by astrogliosis, microgliosis, and altered neuronal morphology with impaired neurogenesis.Histopathology in the brains of DS mice includes aberrant myelination, 25 neuroinflammation, 23 and increased apoptosis. 26These findings support clinical evidence in DS, where white matter damage is evident on MRI 27 and in resected brain sections at autopsy. 28,29White matter damage is increasingly recognized as a common epileptogenic pathology. 30,31Conditional knockout and knockdown studies in DS models suggest that nonseizure comorbidities may occur as primary consequences of the Scn1a mutation in addition to secondary consequences of seizure burden, [32][33][34][35][36][37][38][39][40] confirming clinical reports. 7he clinical data provide incontrovertible evidence that fenfluramine suppresses convulsive seizures in patients with DS. [8][9][10][11] Evidence in clinical studies, 8,13 nonmammalian DS models, 16 and mammalian non-DS models 17 suggests that fenfluramine may also affect phenotypes that are caused by Scn1a haploinsufficiency but not necessarily secondary to seizures.Here, we present the first study to investigate the effect of fenfluramine on myelination, neuroinflammation, and apoptosis in the Scn1a +/− DS mouse brain, as well as survival as an endpoint in the Scn1a +/− DS mouse model.

| Study ethics
All experiments were approved by the Molecular Medicine Research Institute (MMRI; Sunnyvale, CA) Institutional Animal Care and Use Committee (IACUC) and conformed to the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

| Mice
The Scn1a +/− transgenic DS mouse model has been previously described by Kearney et al. 24,41 The heterozygous Dravet mice (Scn1a tm1Kea 50% C57BL/6J, 50% 129S6/SvEvTac background strain) were from The Jackson Laboratory (Bar Harbor, ME, USA).Briefly, a heterozygous Scn1a +/− null allele was generated in TL1 embryonic stem (ES) cells by targeted deletion of the first coding exon (129S6/SvEvTac).These mice were established and maintained as a co-isogenic strain on the 129S6/SvEvTac (129) background.Crossing 129.Scn1a +/− mice to strain B6 resulted in (B6x129) F1.Scn1a +/− (F1.Scn1a +/− ) offspring. 24Genotype was confirmed at postnatal day (PND) 14 and again at harvest by real-time PCR of tail tips at Transnetyx (Cordova, TN) with a custom primer-probe set based upon those applied by Miller et al. 24 Animals were housed in temperature-controlled environments with a 12-hour light/dark cycle and ad libitum access to food and water.Experiments used male and female mice in approximately equal ratios.

| Treatment
Treatment was administered by subcutaneous injection to Scn1a +/− mice once daily from PND 7 through PND 35-37.For comparison, wild-type mice were injected with vehicle or fenfluramine in parallel with Scn1a +/− mice.An injection volume of 10 μL/g body weight was used to deliver dosages of 15 mg/kg racemic fenfluramine (Penn Pharma/ PCI Pharma Services, Tredegar, Gwent, Wales; dissolved in saline) or 0.9% saline solution (vehicle control).PND ages for this study were selected to avoid periods of high demyelination and microglial activity associated with development; normal axonodendritic, axonosomatic, and axonoaxonic synaptic pruning occurs in all mice, with peak activity at PND 21. 42,43 Dose of fenfluramine was chosen based on the lowest reported dose to affect survival in the DBA/1 mouse model (15 mg/kg/day intraperitoneal). 17A dose of 15 mg/kg/day fenfluramine was based on calculations to determine the conversion factor necessary to account for interspecies differences in pharmacokinetics and metabolism (see references 17,44 ).][10][11] Subcutaneous mode of administration was chosen due to the small size of the animals early in the experiment (closer to PND 7) and the associated risk in puncturing and damaging the viscera if using the intraperitoneal mode.

| Tissue preparation
At the conclusion of the dosing period, surviving mice were deeply anesthetized with isoflurane and exsanguinated by cardiac puncture, then formalin-fixed by whole body perfusion using 6 mL ice-cold phosphate buffered saline, followed by 6 mL ice-cold 4% paraformaldehyde.Right brains were harvested and processed as formalin-fixed, paraffin embedded (FFPE) specimens by incubating in 10% neutral buffered formalin at room temperature for 24 hours, transferring to phosphate buffered saline containing 0.02% sodium azide preservative, and embedding in paraffin blocks.FFPE specimens were sectioned into 5-μm sagittal sections by microtome and mounted onto glass slides for immunohistochemistry and imaging studies.Sections within approximately 300 μm of the right-brain midline were used for immunohistochemistry. Nine age-matched wild-type mice were treated with vehicle and served as controls for the histopathology studies.

| Immunohistochemistry
To detect degraded myelin, 5-μm FFPE sagittal right brain sections were immunostained using a 1:1000 dilution of primary degraded myelin basic protein antibody (D-MBP; rabbit-anti-mouse, Millipore Sigma Aldrich D-MBP AB 5864) without antigen retrieval.This antibody preferentially detects debris of myelin degradation without crossreacting to normal myelin.Prior to antibody incubation, sections were neutralized for endogenous peroxidase and phosphatase with a 10-minute incubation with Bloxall (Vector Laboratories SP-6000; Newark, CA), rinsed twice with 0.1% Tween, Tris-buffered saline, and blocked for non-specific binding with 2.5% normal horse serum (Vector Laboratories S-2012-50).Blocking serum was replaced with primary antibody diluted in Antibody Diluent Reagent Solution (003218; Life Technologies), and sections were incubated at room temperature for 1 hour.Slides were washed and incubated 30 minutes with secondary alkaline phosphatase (AP)-conjugated anti-rabbit IgG (Vector ImPress-AP; Vector Laboratories) with Vector Blue AP substrate as a chromagen.Slides were washed and counterstained with non-specific Nuclear Fast Red to provide anatomical context.Slides were dehydrated, cleared, and coverslip-mounted with Cytoseal 60 (Item 8310-4; Thermo Fisher Scientific, South San Francisco, CA) prior to imaging.
To detect activated microglia, 5-μm FFPE sagittal right brain sections were immunostained with 1:12000 dilution of primary microglial activation CD11b antibody (rabbit-anti-mouse; Abcam Item ab133357, Cambridge, MA) with heat-induced antigen retrieval.CD11b antibody identifies inflammatory microglia that phagocytose damaged myelin and represent a secondary marker of myelin damage.Secondary antibody staining and detection were accomplished by the protocol described in the previous paragraph.

| TUNEL assay for apoptosis
Click-iT terminal deoxynucleotidyl transferase-dUTP nick end labeling (TUNEL) was used to detect DNA fragmentation in situ according to the manufacturer's instructions (Item C10617; Thermo Fisher Scientific, South San Francisco, CA). 45Coverslips were mounted using Vectashield Vibrance Antifade hardening mounting medium (H-1700; Vector Laboratories).

| Microscopy
Imaging studies were conducted using an Echo Revolve fluorescence microscope (Echo, San Diego, CA).D-MBP and CD11b immunofluorescence were visualized by the DAPI (UV) excitation/emission channel under 20× magnification.Non-specific Nuclear Fast Red staining was visualized under the Texas Red excitation/emission channel for anatomical context in image overlays.TUNEL-labeled apoptotic nuclei were visualized in the FITC excitation/ emission channel under 4× magnification fields stitched together with Affinity Photo (Serif, Nottingham, UK).

| Image quantification
Images were quantified manually using Image J software (v1.53qNational Institutes of Health, Bethesda, MD) within the regions of interest for each mouse (unblinded to genotype and treatment; performed by a single investigator).D-MBP staining in the parietal cortex and hippocampus was quantitatively scored.Quantification included both a 0.1 mm 2 area of the CA3 region of the hippocampus, including the stratum pyramidale (SP), stratum oriens (SO), stratum lucidum (SL), and stratum radiatum (SR), and a 0.1 mm 2 area of the adjacent parietal cortex above the corpus callosum for each mouse.For activated microglia expressing CD11b, a 20× field of the genu of the corpus callosum and the perforant pathway of the hippocampus was scored for each mouse.CD11b + microglia in the corpus callosum and hippocampus were scored semi-quantitatively on a 0-5 Likert scale (0, no activated microglia; 1, low numbers of activated microglia; 2, noticeably increased activated microglia; 3, intermediate activated microglia; 4, highly activated microglia; 5, very highly activated microglia).Scores in the genu of the corpus callosum and perforant pathway of the hippocampus were summed, for a maximum possible score of 10.Activated microglial cells were identified by unambiguous increase in microglial body size, ramification thickness, length (in rod-shaped, activated microglial cells), clear phagosome extension, or amoeboid change in shape.TUNEL-labeled apoptotic nuclei were quantified by manual counting of positive nuclei in Image J (v1.53q).

| Survival monitoring
We maintained the Scn1a +/− mice and evaluated survival for an extended period of time in untreated animals (n = 99) before we injected treatment groups for the histological experiments.The untreated mice and all treatment groups were housed and maintained under identical conditions.We monitored and documented survival daily from PND 7 to PND 35-37.Occasionally, we video-recorded 3 Scn1a +/− mice using wide-angle infrared cameras.

| Statistical assessment
Survival was computed from PND 7 to PND 35-37 or death.Quantitative differences in levels of the histopathological markers D-MBP, CD11b +, and TUNEL among treatment groups were assessed using a 1-way analysis of variance (ANOVA) with a post-hoc Dunnett's test for multiple comparisons.Dunnett's tests were versus the vehicle-treated DS mice control group.The survival statistics were evaluated using all Scn1a +/− mice, including untreated mice that were maintained in our laboratories earlier.Survival distributions of treatment groups were described using Kaplan-Meier methods, and the Logrank Mantel-Cox test was used to assess differences in survival between treatment groups.P-values <0.05 were considered statistically significant.

| Diazepam treatment
In supplementary studies, 24 Scn1a +/− mice were treated with 10 mg/kg diazepam per Tiraboschi et al. 16 (see Data S1).Survival was evaluated and compared with fenfluramine-or vehicle-treated Scn1a +/− mice, untreated Scn1a +/− mice, and wild-type controls.Immunohistochemistry with D-MBP or CD11b antibodies and TUNEL assays for apoptosis were also conducted in surviving animals as described above and in the Data S1.

| Demyelination
FFPE sections for histopathology analysis were prepared for DS mice that completed treatment through PND 35-37 in the survival study (vehicle, n = 9; fenfluramine, n = 13).Nine sections were prepared in parallel for wild-type controls.Degenerated myelin was assessed in brain sections from DS mice with D-MBP staining and scoring under visible light and fluorescence microscopy (Figure 1).Representative sections demonstrated evidence for increased myelin damage in the hippocampus, as indicated by a qualitative increase in spherical blue dots in DS mice relative to wild-type controls (Figure 1A).In general, these sections showed less visibly intact linear myelin cross-detected by the D-MBP antibody than wildtype controls.
Overall, microscopic evaluation of the sagittal sections revealed that the regions with the most D-MBP immunofluorescence were the hippocampus and parietal cortex (representative images, Figure 1B,C, respectively).In these regions, spheroidal and punctate debris was detected by the D-MBP antibody.For quantitative image analysis, these regions were combined.
Image quantification of myelin debris per unit area (0.1 mm 2 of the CA3 region of the hippocampus and 0.1 mm 2 parietal cortex) demonstrated that the level of degenerated myelin was significantly higher in the DS mice (818 ± 101; ANOVA, P < 0.0001) compared to wild-type mice (575 ± 53, P < 0.0001 by Dunnett's post-hoc test for multiple comparisons; Figure 2).Treatment with fenfluramine significantly reduced the level of damaged myelin in DS mice (635 ± 101) compared to vehicle-treated DS mice (P = 0.0002 by Dunnett's test; Figure 2).

| Activated microglia
Activated CD11b + microglia were more apparent in the genu of the corpus callosum and the perforant pathway of the hippocampus (Figure 3A,B, respectively) than any other brain region.Sections from the vehicle-treated DS mice were enriched in clearly elongated, rod-shaped microglia with clear phagosome extension (central panel; note arrowheads) compared to wild-type mice (left panels).
DS mice exhibited spontaneous seizures.Video recordings were available from three DS mouse controls (Data S1, Videos S1-S3); all three mice died immediately following a convulsive seizure rated 5 on the Racine seizure rating scale. 47Body position at time of death showed hindlimb extension at 180 degrees (see reference 48 ) in all three recordings.

| DISCUSSION
This study is the first to demonstrate that chronic treatment with fenfluramine in DS mice (a) reduces levels of myelin degeneration, microglia activation/neuroinflammation, and apoptosis and (b) increases survival.The effects of fenfluramine were most pronounced in the parietal cortex and CA3 of the hippocampus for demyelination, in the corpus callosum (genu) and perforant pathway of the hippocampus for activated microglia, and in the corpus callosum for apoptosis.

| Fenfluramine impact on myelin degradation
Our study provides the first evidence in a DS mouse model that fenfluramine reduces neuronal markers of myelin damage.Atypical myelinogenesis in the corpus callosum was evident before seizure onset in another Scn1a mouse model of DS. 25 White matter (myelin) damage is a common pathology in animal models of epilepsy and in patients with epilepsy and persons with demyelinating motor diseases (e.g., multiple sclerosis). 30,31,49urther, the spheroidal D-MBP staining we observed in the DS mouse hippocampus is consistent with clinical studies in patients with DS showing elevated white matter damage throughout the brain by MRI 27,50 and upon autopsy. 29t is unclear whether myelination protection/repair is a secondary consequence of reduced seizure activity by fenfluramine, whether fenfluramine acts through a direct mechanism that either protects or repairs myelin, or a combination of both.Myelin damage is a general consequence of epilepsy. 28,29However, fenfluramine's dualaction, serotonergic and sigma-1 receptor mechanism of action is unique among antiseizure medications. 51A growing body of evidence supports a combination of serotonergic and Sigma1R positive modulatory activity in pro-survival, pro-cognitive effects of fenfluramine in addition to control. 513][54] Sigma1R was identified as a critical regulator of inflammation in a knockout Sigma1R mouse model, 52 and serotonin reduced levels of inflammatory cytokines in brain homogenates of rats after pentylenetetrazoleinduced seizures. 55Sigma1R regulates myelination in the brain, with defects associated with neurodegenerative disorders (e.g., multiple sclerosis). 53,54,56In a mouse model of spatial learning and memory, fenfluramine positively modulated Sigma1R to prevent dizocilpine-induced amnesia, 15 and later reports suggested that the neuro (active) steroids dehydroepiandrosterone sulfate and pregnenolone sulfate were endogenous ligands mediating these neuroprotective effects. 14Activation of Sigma1R by the neurosteroid dehydroepiandrosterone contributed to hippocampal neurogenesis in neurodegenerative mouse models. 56Whether the dual-action serotonergic/Sigma1R pharmacology of fenfluramine contributes to preventing myelin damage in DS mice remains to be established.

| Activated microglia
To our knowledge, this is the first report of fenfluramine decreasing activated CD11b + microglia in a mammalian DS model.CD11b + microglia engulf and recycle cellular debris, including degraded myelin.Reduction of CD11b + immunofluorescence after fenfluramine may reflect preserved myelin integrity.Moreover, CD11b immunostaining was specifically selected because it is a marker of active microglia with a phagocytic phenotype. 57CD11b detects neuroinflammation.A reduction in CD11b + microglia suggests reduced neuroinflammation and myelin damage after fenfluramine treatment.It is unclear whether these effects are a result of reduced seizure activity or a direct result of fenfluramine on neuroinflammation in these brain regions.The differences observed in the neuroanatomical location of the fenfluramine effect on demyelination (CA3 of hippocampus and parietal cortex) and microglia (perforant pathway of hippocampus and genu of corpus callosum) also remain to be investigated.Neuroinflammation in context of oxidative stress has been described immunohistochemistry studies of resected brain tissue of patients with epileptic cortical malformations, including tuberous sclerosis complex (cortical tubers), hemimegalencephaly, and focal cortical dysplasia. 58Further, white matter myelin damage was reported by myelin immunohistochemistry in resected brain tissue from patients with tuberous sclerosis complex or focal cortical dysplasia. 59The functional consequences of reduced neuroinflammation after fenfluramine remain to be investigated.

| Apoptosis
The observation that fenfluramine inhibited apoptosis in DS mice is a novel finding, but the implications remain to be established.There is some evidence that apoptosis impacts the excitatory/inhibitory imbalance in both seizures and non-seizure comorbidities. 26,46A reduction in pathological apoptosis long-term may afford better seizure control and improved cognition, as demonstrated after administration of liraglutide in an Scn1a knockout model. 26Apoptotic neurodegeneration has also been cited as a mechanism for worsening cognition after treatment of various antiseizure medications. 60Patients with DS are typically taking multiple concomitant antiseizure medications. 61Some antiseizure medications inhibit endogenous neuroprotection mechanisms in the brain, resulting in reduced brain mass and cognitive impairment.Fenfluramine, by contrast, improved the regulation of behavior, cognition, and emotions in clinical studies in patients with DS. 13,62 The amelioration of apoptosis in the corpus callosum in our study by fenfluramine could be a result of restoring neuroprotective mechanisms, such as neurotrophin systems critical for neuronal survival. 60Alternatively, apoptosis may play a role in structural remodeling of the CNS architecture in specific brain regions.Prior reports have demonstrated the importance of remodeling of neurons, astrocytes, and glia in impacting neuronal survival, pruning, and dendritic outgrowth. 63This hypothesis has some support in a zebrafish scn1lab mut/ mut DS model where fenfluramine preserved the architectural integrity of inhibitory GABAergic neurons. 16urther experiments are needed to determine whether the impact of fenfluramine on apoptosis in the corpus callosum plays a role in neuroprotection, neuronal remodeling, anti-inflammation, and/or a combination of these and other mechanisms.

| Survival
Our study provides the first evidence that fenfluramine promotes survival in a DS mouse model.The enhanced survival with fenfluramine in DS mice is consistent with a report of lower all-cause mortality and SUDEP rates (both rates were 1.7 per 1000 person-years in 732 patients with DS who were treated with fenfluramine over a total of 1185.3 person-years of exposure) compared with a published cohort study in patients who were not receiving fenfluramine 64 (all-cause mortality rates of 15.8 deaths per 1000 person-years and SUDEP rates of 9.3 per 1000 person-years over a total of 1073 person-years). 64Further experimentation is required to determine whether the improved survival after fenfluramine we observed in our study is linked to reduction in SUDEP.Evidence from an audiogenic seizure DBA/1 mouse model showed that fenfluramine treatment reduced risk of convulsive seizure-induced respiratory arrest (S-IRA) preceding SUDEP at concentrations which had no impact on total seizure threshold (15 mg/kg fenfluramine for 8 h). 17The effect of fenfluramine was more potent at blocking S-IRA than seizures at this dose, and was inhibited serotonergic 5-HT 4 receptor antago- 21 The authors concluded that seizures and IRA likely followed related but non-overlapping mechanisms, and that fenfluramine acted pharmacologically on both pathways.In prior studies with DS mice, SUDEP occurs immediately following convulsive seizures, and conditional knockout of Scn1a in brain recapitulated the SUDEP phenotype. 36Although the three videos of the animals that died in our study are consistent with a convulsive seizure prior to sudden death, video recording of all animals is needed for definitive proof.

| Limitations
This study has some limitations.First, seizure activity was not quantified in this study, nor was continuous, 24-hour video surveillance conducted on all animals.Our study reports all-cause mortality; additional studies would be needed using electroencephalogram (EEG) and continuous video surveillance on all study animals to determine whether all deaths were preceded by a convulsive seizure, as well as the impact of fenfluramine on the number, intensity, and duration of seizures.Second, the fenfluramine dose chosen (15 mg/kg/ day) prevented S-IRA preceding SUDEP in the DBA/1 mouse model without significantly affecting the seizure threshold. 17Additional experiments are needed to determine whether fenfluramine also inhibits S-IRA in the DS Scn1a +/− mouse model, and to determine the correlation between the pro-survival effect of 15 mg/ kg/day fenfluramine and both seizure threshold and neuro-histopathology. Third, the neuroimaging results reported in this study are limited by survivor bias.Additional studies are needed to evaluate neural histopathology immediately after death to determine whether histopathology differed in the mice that died compared to survivors.Finally, our study reports the use of fenfluramine as monotherapy, whereas clinical settings use an add-on medication to existing further studies are needed to determine the effect of fenfluramine in combination with routinely used antiseizure medications on survival and markers of neuroinflammation in mouse models of DS.

| Future directions
Further studies are warranted to confirm and extend our neuroinflammation results.Electron microscopy is needed to quantify G-ratio to better assess whether demyelination is ongoing in Dravet mice, and to assess the effects of fenfluramine on this endpoint.Highmagnification, high-resolution confocal microscopy images can better assess morphological changes in CD11b + activated microglial cells.Additional apoptosis markers are needed to corroborate the TUNEL staining (e.g., Bax, Bcl2, cleaved caspase), with high-resolution confocal imaging to identify apoptotic cell types using double immunofluorescence (e.g., main neuronal/glial cell types and apoptotic markers).
Our survival results warrant future studies to investigate the mechanisms behind the observed pro-survival outcome of fenfluramine treatment.Most importantly, obtaining seizure counts (number and intensity) and EEG recordings is an important next step in understanding the mechanism behind how 15 mg/kg/day fenfluramine improves survival in Scn1a +/− mice.Recently published studies investigating outcomes of novel antiseizure medications in the Scn1a +/− mouse model provide the experimental framework for future studies (see Hawkins et al. 65 for a recent example), including behavior testing, pharmacokinetics to confirm brain and plasma exposure and compare subcutaneous versus intraperitoneal routes of administration, continuous video surveillance with EEG recordings, and Racine scoring for seizures. 17Our study is the first to report fenfluramine treatment of DS mice reduces neuroinflammation demyelination-most notably in the hippocampus, corpus callosum, and parietal cortex-and increases survival.This study provides potential neuropathological pathways for the impact of fenfluramine on promoting survival and reducing neuroinflammation, and is hypothesis-generating for insights into the neuroanatomical structure-function relationships.Further studies are needed to determine whether the histopathological changes following fenfluramine are related to decrease in SUDEP-related mortality and whether pro-survival is solely secondary to seizure control, or whether additional mechanisms are involved.

F I G U R E 2
Damaged myelin quantitation in parietal cortex and hippocampus CA3.Myelin debris was quantified by fluorescence microscopy in 5-μm sagittal brain sections via image quantification with D-MBP antibody immunostaining.Data are plotted as mean ± SD.P-values were calculated by ANOVA with post-hoc Dunnett's test for multiple comparison.ANOVA results: P < 0.0001.Dunnett's multiple comparisons test results compared with DS mice, vehicle-treated group: ***P < 0.001 (left to right, vs DS Mice Vehicle: WT Mice Vehicle, P < 0.0001; DS mice FFA, P = 0.0002).ANOVA, analysis of variance; DS mice, Scn1a +/− mouse model of Dravet syndrome; FFA, fenfluramine; WT, wild-type.

F I G U R E 4
Image quantification of the effect of fenfluramine treatment on CD11b + immunostaining of microglia in the hippocampus and corpus callosum.Images (one 5-μm sagittal section per mouse) were evaluated semi-quantitatively on a Likert scale of 0-5 (0, no activated microglia; 5, highly activated microglia).Data are plotted as mean ± SD; P-values were calculated by ANOVA with post-hoc Dunnett's test for multiple comparison.ANOVA results: P = 0.0040.Dunnett's multiple comparisons test results compared with DS mice, vehicle-treated group: **P < 0.01; *P < 0.05 (left to right, vs DS Mice Vehicle: WT Mice Vehicle, P = 0.0022; DS mice FFA, P = 0.0317).ANOVA, analysis of variance; DS mice, Scn1a +/− Dravet syndrome mice; FFA, fenfluramine; WT, wild-type.

F I G U R E 5
Apoptosis by TUNEL staining in 5-μm sagittal sections of the corpus callosum in wild-type mice or surviving DS mice treated subcutaneously once daily with vehicle (n = 9) or 15 mg/kg fenfluramine (n = 13) from PND 7 to PND 35-37.A, Representative images from genu of the corpus callosum (scale bar, 130 μm; green [FITC] channel, 10× objective) and (B) manual image quantification of apoptotic nuclei by TUNEL using 4× magnification fields stitched together with Affinity Photo.In B, data are plotted as mean ± SD; P-values were calculated by ANOVA with post-hoc Dunnett's test for multiple comparison.ANOVA results: P = 0.0087.Dunnett's multiple comparisons test results compared with DS vehicle-treated group: **P < 0.01 (left to right, vs DS Mice Vehicle: WT Mice Vehicle, P = 0.0049; DS mice FFA, P = 0.0543).ANOVA, analysis of variance; DS mice, Scn1a +/− Dravet syndrome mice; FFA, fenfluramine; PND, post-natal day; TUNEL, terminal deoxynucleotidyl transferase biotin-dUTP nick end labeling; WT, wild-type.