The association of serotonin reuptake inhibitors and benzodiazepines with ictal central apnea

SUMMARY

Objective

Ictal (ICA) and post-convulsive central apnea (PCCA) have been implicated in sudden unexpected death in epilepsy (SUDEP) pathomechanisms. Previous studies suggest that serotonin reuptake inhibitors (SRIs) and benzodiazepines (BZDs) may influence breathing. The aim of this study was to investigate if chronic use of these drugs alters central apnea occurrence in patients with epilepsy.

Methods

Patients with epilepsy admitted to Epilepsy Monitoring Units in nine centers participating in a SUDEP study were consented. Polygraphic physiological parameters were analyzed, including VEEG, thoraco-abdominal excursions and pulse oximetry. Outpatient medication details were collected. Patients and seizures were divided into SRI, BZD and control (no SRI or BZD) groups. ICA and PCCA, hypoxemia and electroclinical features were assessed for each group.

Results

476 seizures were analyzed (204 patients). The relative risk for ICA in the SRI group was half that of the control group (p=0.02). In the BZD group, ICA duration was significantly shorter than in the control group (p=0.02), as was post-ictal generalized EEG suppression (PGES) duration (p=0.021). Both, SRI and BZD groups were associated with smaller seizure-associated oxygen desaturation (p=0.009; p<0.001). Neither presence nor duration of PCCA were significantly associated with SRI or BZD (p>0.05).

Conclusions

Seizures in patients taking SRIs have lower occurrence of ICA and patients on chronic treatment with BZDs have shorter ICA and PGES durations. Preventing or shortening ICA duration by using SRIs and/or BZD in patients with epilepsy may play a possible role in SUDEP risk-reduction.

INTRODUCTION

Sudden unexpected death in epilepsy (SUDEP) is second only to stroke as a neurological cause of total years of potential life lost in the United States [1]. Video-EEG monitored SUDEP cases suggest breathing dysfunction following generalized tonic-clonic seizures is involved in mechanisms of death, and terminal post-convulsive central apnea (PCCA) precedes asystole [2]. These observations, together with imaging [3] and neuropathological [4] evidence, suggest both functional and structural compromise of brainstem function. Some SUDEP and near SUDEP cases occur with focal seizures without secondary generalization [5]. Ictal central apnea (ICA) is frequently seen during focal seizures (37–44%) and can be prolonged (> 60 seconds) [6]. Prolonged ICA may predispose to fatal outcomes, and has been proposed as a potential biomarker of SUDEP [7, 8].

Serotonin (5-HT, 5-hydroxytryptamine) is a neurotransmitter important to chemoreception (CO₂ and pH) [9], and arousal [10], both of which may be compromised in SUDEP [11]^,[12]. Serotonergic dysfunction has been posited as central to both SUDEP and Sudden Infant Death Syndrome [13–15] mechanisms, and SUDEP animal models of serotonin deficiency lend support to this hypothesis [16]. Post-ictal apneic deaths in one such model (DBA/2 mice) can be prevented by pretreatment with serotonin re-uptake inhibitors (SRIs), probably through promotion of respiratory rhythmogenesis, rather than increase in ventilatory drive [17, 18]. Human evidence is scant; there is only one study which found that focal onset seizures in patients on SRIs have less frequent hypoxemia (SpO₂< 85%), but whether SRIs do so through decrease in peri-ictal apnea incidence or duration is unknown or unconfirmed [19]. Gamma-aminobutyric acid (GABA), the major inhibitory central nervous system neurotransmitter, critically regulates respiratory rhythmogenesis and is required for normal respiratory motor pattern generation [20]. Benzodiazepines (BZDs) are GABA_A receptor positive allosteric modulators and efficacious antiepileptic drugs [21, 22]. Therapeutic doses have minimal breathing suppression effects if any, but in excess or in conjunction with opiates, they may cause breathing dysfunction[23] and respiratory arrest.[24] Animal and human studies of chronic benzodiazepine use at therapeutic doses have demonstrated stimulant, probably tachypneic, breathing responses [25, 26]. A recent study has shown that ictal central apnea (ICA) duration is associated with the presence of post-convulsive central apnea (PCCA) [27]. We hypothesized that peri-ictal central apnea can be prevented by chronic treatment with SRIs and/or BZDs. We investigated the presence and duration of ICA and PCCA, hypoxemia and electroclinical features in seizures in patients being treated with and without SRIs, as well as with and without BZDs.

METHODS

Patients and clinical settings.

We prospectively studied 476 seizures in 204 patients. Of these, 312 seizures were part of a study of ICA incidence[6]; 148 generalized clonic seizures (GCS) were reported in another partially overlapping study that analyzed ICA and PCCA incidences [27, 28] All were consented participants in the NINDS Center for SUDEP Research’s Autonomic and Imaging Biomarkers of SUDEP multi-center project (U01-NS090407) and its precursor, the Prevention and Risk Identification of SUDEP Mortality (PRISM) Project (P20NS076965).

Patients aged ≥16 years with intractable epilepsy (failure of adequate trials of at least two antiepileptic medications) [29], who were undergoing video-electroencephalography (VEEG) evaluation in the epilepsy monitoring units (EMUs) of participating centers from September 2011 until April 2018 were studied. Inclusion criteria included patients with thoraco-abdominal belts during recorded seizures in the EMU. Exclusion criteria were status epilepticus, and unavailable plethysmography or video during recorded seizures. Demographic and clinical data were collected, including epilepsy duration, epilepsy syndrome and awake or sleep states at seizure onset. Seizures were divided into two groups: generalized convulsive (GCS) and non-convulsive (NCS) seizures.

Serotonin reuptake inhibitors (SRIs) and Benzodiazepines (BZDs)

Regular outpatient medication details were collected during EMU admissions in all patients. Information regarding treatment with SRIs and BZDs was collected. SRIs included selective serotonin reuptake inhibitors (SSRIs) and serotonin–norepinephrine reuptake inhibitors (SNRIs). Only patients chronically treated with BZDs were included, including those whose home medications were discontinued during admission. Patients acutely administered BZDs or treated with opioids were excluded from analysis in order to prevent a confounding influence on breathing metrics [30, 31]. Analysis of SRIs and BZDs was performed separately; patients were divided into: SRI, BZD, control group (no SRI or BZD) or both (SRI/BZD). Patients on both, SRIs and BZDs, were excluded from SRI and BZD groups to avoid confounding bias.

Cardiorespiratory monitoring and VEEG monitoring

All patients had prolonged surface VEEG monitoring using the 10–20 International Electrode System. Electroencephalography (EEG) and electrocardiogram (EKG) were acquired using Nihon Kohden (Tokyo, Japan), Micromed (Modigliani Veneto, Italy) and Xltek (Natus) acquisition platforms. Peripheral capillary oxygen saturation (SpO₂) was monitored using pulse oximetry (Nellcor OxiMax N-600x [Convidien], Masimo Radical-7 [Irvine] and SenTec Digital Monitoring System [Therwil BL]) and plethysmography (Ambu [Ballerup, Denmark] Sleepmate and Perfect Fit 2 [Dymedix]). Chest wall and abdominal excursions were recorded using inductance plethysmography (Ambu, Ballerup, Denmark and Sleepmate or Perfect Fit 2, Dymedix, St Paul, MN, USA).

We defined central apnea as breathing cessation of ≥5 seconds without alternative explanation (i.e. speech, movement or intervention). This definition is based on minimum apnea period induced by cortical stimulation[32], and is different to that used in sleep studies. In sleep studies, a minimum apnea period of 10 seconds has been a constant definition. On the other hand, in epilepsy studies, ICA definition has varied between different studies from one or two missed breaths [28, 33] to 10 or 15 seconds [6, 7, 34] More recently, it has been found that using ≥5 seconds may be a more accurate definition during seizures, and using the sleep definition of apnea may underestimate this phenomenon in epilepsy. This conclusion is based on stimulation studies aiming to identify the symptomatogenic zone of ICA [32]. They have demonstrated that stimulation of such areas typically induces a minimum apnea period of 5 seconds, even during short stimulation sessions of 2 seconds [32]. ICA was assessed in NCS and in the pre-convulsive phase in GCS, either in focal or generalized epilepsies, since in the generalized epilepsies, absence, myoclonic or electrographic seizures can occur before the convulsive phase. ICA was not evaluated during GCS, because of invariable artifact presence in breathing channels due to generalized tonic posturing or clonic movements. Post-convulsive central apnea (PCCA) referred to apnea that occurred after GCS. Breathing evaluation was made through careful composite analysis of inductance plethysmography, video recording and SpO₂. Baseline hypoxemia was defined two minutes pre-ictally as mean SpO₂ in a 15 second, artifact free epoch. For GCS and NCS, the overall desaturation nadir referred to the lowest SpO₂ value registered during and up to 3 minutes after the seizure. To evaluate respiration in the pre-convulsive phase of GCS, an additional desaturation nadir was considered during this phase.

Seizure duration was defined as the period from electrographic seizure onset to end. Duration of convulsive phase was defined as the period from onset of motor generalization to the end of the last clonic movement. Presence and duration of postictal generalized EEG suppression (PGES)[35] was analyzed by a validated automated EEG suppression detection tool[36].

Statistical analysis.

Statistical analysis was performed using the Statistical Package for Social Science (SPSS, version 24; IBM Corp, Armonk, NY, USA). Summary statistics were reported as mean ± standard deviation (median, range). Relative risk (RR) for all the variables at a seizure level was assessed by Generalized Estimating Equation (GEE) with same subject exchangeable correlation to take into account the variable number of seizures per patient. Nominal patient characteristics were compared across medication groups with Pearson Chi-Square or Fisher exact tests. Continuous characteristics were compared by Mann-Whitney U tests or Spearman correlations. Binary logistic regression was used to assess associations between dichotomous variables SRI (+/−) or BZD (+/−) with other variables. A p-value <0.05 was considered statistically significant.

RESULTS

General description

A total of 476 seizures were analyzed (204 patients, 106 females [52%]). Mean age was 39.2 years ± 14.5 (36; 17–77). Mean epilepsy duration was 18.3 years ± 13 (18; 0–52). Epilepsy was classified as generalized in 28/204 (13.7%) patients (45 seizures), and focal in 173/204 (84.8%) patients (426 seizures). One patient (2 seizures) had both generalized and focal, and two (3 seizures) had indeterminate epilepsy. Seizures were classified as GCS (231/476 [48.5%] seizures in 126 patients) or NCS (245/476 [51.5%] seizures in 96 patients). Eighteen patients had both GCS and NCS. Two hundred and thirty-seven seizures (49,8%) occurred during wakefulness and 239 (50,2%) out of sleep. PGES was present after 164/238 (69%) GCS, with a mean duration of 39.3 seconds ± 22.1 (37; 1–169). No PGES was seen after NCS.

Thirty-four patients were on SRIs: 30 on SSRIs: escitalopram (9), citalopram (6), sertraline (5), vilazodone (2), fluoxetine (6) and paroxetine (2), 6 patients on SNRIs: duloxetine (2), venlafaxine (1) and amitriptyline (3). Two patients were on both citalopram (SSRI) and duloxetine (SNRI). Twenty-six patients were on benzodiazepines (clobazam [11], clonazepam [11], lorazepam [2] and alprazolam [2]). Six patients were concurrently treated with BZDs and SRIs and were excluded from the analysis of either drug.

Peri-ictal central apnea

In 69/476 (14.5%) seizures, ICA could not be commented upon due to movement or acquisition artifact and/or obstructed video. In the remaining 407 seizures in 187 patients, ICA was present in 149/407 (36.6%) seizures in 77 patients (39.6%) and exclusively seen in focal epilepsy (p<0.0001). Mean ICA duration was 22.8 seconds ± 17.2 (17; 5–97). Of 231 GCS, seven seizures in seven patients were excluded from PCCA analysis because of post-ictal artifact. PCCA was detected in 39/224 (17.4%) GCS in 28/119 patients (23.5%) with focal or generalized epilepsy. Mean PCCA duration was 11.7 seconds ± 12.9 (8; 5–85).

A). Serotonin Reuptake Inhibitors (SRIs) analysis SRIs and Ictal Central Apnea (ICA).

The SRI was compared to the control group. Reliable inductance plethysmography and unobstructed video for ICA diagnosis was available in 315 seizures in 151 patients (Figure 1). In the SRI group, ICA was seen in 15/78 (19.2%) seizures, whereas in the control group, it occurred in 112/237 (47.2%), p<0.001 (Figure 2 A ). The relative risk for ICA was double in the control group compared to SRI (RR 2.013, CI 95% 1.071–3.783, p=0.02). Mean apnea duration did not differ between SRI (27.6 seconds ± 22.9) and control groups (23.8 seconds ±17.6) groups (p=0.7).

Figure 1. Flowchart for study population.

SRI: serotonin reuptake inhibitor, BZD: benzodiazepine, GCS: generalized convulsive seizures, NCS: non-convulsive seizures, ICA: ictal central apnea, PCCA: post-convulsive central apnea. *Thoraco-abdominal excursions signal not obscured by movement and/or acquisition artifact and unobstructed video.

Figure 2. SRI and BZD influences on peri-ictal respiration.

A and B) Bar plots for data from non-convulsive seizures (NCS) grouped with the pre-convulsive phase of generalized convulsive seizures (GCS), showing the association between A) SRI and ICA incidence and B) SRI and SpO₂ drop (percentage). In the BZD group, C) ICA duration was significantly shorter and D) SpO₂ drops were smaller than in control. There was a positive correlation between ICA duration and SpO₂ drop as shown by a robust simple linear regression line and 95% confidence intervals. For parts A-C values are mean +/− SEM.

ICA: ictal central apnea, No: number, SpO₂: peripheral capillary oxygen saturation, SRI: serotonin reuptake inhibitor, BZD: benzodiazepine.

There were no significant differences between SRI and control in gender (p=0.7), age (p=0.2), epilepsy duration (p=0.6), awake/sleep state (p=1), type of epilepsy (p=0.7), epileptogenic zone (p=0.3), oxygen administration (p=0.8) and airway suction (p=0.6) after GCS.

SRIs and Post-Convulsive Central Apnea (PCCA).

In the SRI group, 7/36 (19.4%) GCS had PCCA whereas in the control group, PCCA was seen in 27/162 (16.6%) GCS (Figure 1). Neither presence (p=0.2) nor duration (p=0.9) of PCCA were significantly associated with SRI.

SRIs and hypoxemia.

SRI group had smaller seizure-related oxygen desaturations (Table 1; p=0.009). To determine the phase of the seizure in which SRIs may have an effect, we looked at seizures that remained focal (NCS) and the focal phase of seizures that progressed to convulsions together, and found smaller oxygen decreases in this combined analysis (p=0.02) [Figure 2 B]. And then, we looked at the convulsive phase separately, and confirmed that there were no significant differences between SRI and control regarding oxygen decrease in the convulsive post-convulsive phase compared to baseline (p=0.7).

Table 1.

Respiratory and electroclinical findings in SRI, BZD and control groups.

	Control		SRI		p-value	BZD		p-value
ICA incidence (%)	112/237 (47.2%)		15/78 (19.2%)		<0.001 *	22/92 (23.9%)		0.130
PCCA incidence (%)	27/162 (16.6%)		7/36 (19.4%)		0.470	5/26 (19.2%)		0.314
	Mean	SD	Mean	SD	p-value	Mean	SD	p-value
ICA duration (s)	23.8	17.6	27.6	22.9	0.770	14.4	7.2	0.021 *
PCCA duration (s)	12.8	15.2	9.6	4.9	0.906	8.40	2.7	0.642
SpO2 drop (%):
For all seizures	20.5	19.5	12.6	16.7	0.009 *	10.6	17.5	<0.001 *
For NCS and pre- convulsive phase of GCS	4.8	7.7	2.6	5.5	0.023 *	1.7	3.4	0.004 *
For convulsive/post- convulsive phase of GCS	31.6	16.9	32.8	13.2	0.756	31.1	13.7	0.916
Seizure duration (s):
NCS	72.7	47.2	104.4	121.1	0.357	65.1	48.6	0.103
GCS	126.1	55.9	153.4	136.9	0.392	101.1	96.5	0.057
Convulsive phase	59.0	25.0	59.0	24.9	0.121	43.5	14.0	0.016 *
PGES duration (s)	40.2	22.0	39.5	22.0	0.735	28.8	18.3	0.021 *

^*Statistically significant difference, SRI: serotonin reuptake inhibitor, BZD: benzodiazepine, ICA: ictal central apnea, PCCA: post-convulsive central apnea, GCS: generalized convulsive seizures, NCS: non-convulsive seizures.

Electroclinical seizure features.

PGES was observed in 28/37 (75.7%) SRI and in 115/166 (69.3%) control (p=0.5) groups. No difference between groups was observed in PGES duration (p=0.1), convulsive phase duration (p=0.1), electrographic GCS duration (p=0.3) or electrographic NCS duration (p=0.9).

B). Benzodiazepines (BZDs) analysis

BZDs and Ictal Central Apnea (ICA).

Reliable breathing assessment for ICA was available in 329 seizures in 148 patients from the BZD and control groups (Figure 1). In the BZD group, ICA was seen in 22/92 (23.9%) seizures, whereas in the control group, it occurred in 112/237 (47.2%). There were no differences regarding the relative risk for ICA occurrence when comparing BZD vs control (p=0.39), after taking into account the variable number of seizures per patient.

ICA longer duration was inversely associated BZD (p=0.02) [Figure 2 C]. There were no significant differences between BZD and control groups in gender (p=0.6), age (p=0.6), epilepsy duration (p=0.2), awake/sleep state (p=0.9), type of epilepsy (p=0.4), epileptogenic zone (p=0.4), oxygen administration (p=1.0) or airway suction (p=0.8) after GCS.

BZDs and Post-Convulsive Central Apnea (PCCA).

PCCA was found in 5/26 (19.2%) BDZ and in 27/162 (16.6%) control group (Figure 1). Neither presence (p=0.3) nor duration (p=0.6) of PCCA were significantly associated with BZD.

BZDs and hypoxemia.

BZD group had smaller seizure-associated oxygen desaturation (p<0.001) [Table 1&Figure 2D]. To determine the phase of the seizure in which SRIs may have an effect, we looked at both seizures that remained focal (NCS) and the focal (NCS) phase of seizures that progressed to convulsions, and found smaller oxygen decreases in this combined analysis (p=0.004). And then, we looked at the convulsive phase separately, and confirmed that there were no significant differences between BZD and control group regarding oxygen decrease in the convulsive/post-convulsive phase (p=0.9).

When absolute SpO₂ nadirs were analyzed, there was a significant difference between BZD and control groups in the pre-convulsive (p=0.005), but not in the convulsive/post-convulsive phase (p=0.9).

Electroclinical seizure features.

PGES was present in 18/28 (64.3%) BZD group and 115/166 (69.3%) control GCS (p=0.6). There were no statistical differences regarding the RR of PGES when comparing BZD vs control (p=0.6). However, BZD group was significantly associated with shorter PGES duration (p=0.02), shorter convulsive phase duration (p=0.016) and a trend toward shorter seizure duration (p=0.05). On the contrary, in NCS no difference was found in seizure duration (p=0.1).

Based on these findings, we compared the group on both SRIs and BZDs (SRI/BZD) with SRI, and BZD groups, to assess a possible additive effect upon the ICA. In the SRI/BZD group, ICA was seen in 4/10 (40%) seizures. There were no significant differences between SRI/BZD and SRI (p=0.226) or BZD groups (p=0.463) in ICA incidence. Similarly, mean apnea duration did not differ between SRI/BZD (21.5 seconds ±1.3) and SRI (27.6 seconds ± 22.9) [p=0.705] or BZD (14.5 seconds ±7.2) [p=0.059] groups.

DISCUSSION

Our findings suggest that SRIs and BZDs may influence ictal central apnea (ICA) and electroclinical seizure characteristics. Seizures in the SRI group were associated with lower ICA incidence and smaller oxygen desaturations, but they were not related to ICA duration, seizure duration, PGES or post-convulsive central apnea (PCCA). On the other hand, BZDs were associated with shorter ICA duration and smaller oxygen desaturations, but not with ICA incidence. In addition, in the BDZ group, generalized convulsive seizures had shorter duration, as well as shorter post-ictal generalized EEG suppression durations (PGES). In effect, both classes of drugs seem to influence ICA, and further, BZDs were also associated with reduced seizure severity, in terms of shorter GCS and PGES.

ICA may be benign in the majority of cases, but may pose danger when prolonged [6–8], and has been linked to SUDEP and near-SUDEP cases [7, 8]. Stimulation studies have shown that seizure spread to the limbic/paralimbic mesial temporal lobe structures is the most likely mechanism for inhibition or disruption of brainstem respiratory neuronal function in the medulla [37–39]. PCCA appears to be a more deleterious phenomenon, and has been linked to SUDEP and near-SUDEP cases [2, 28]. Such breathing dysfunction propensity may result from longstanding injury to critical brainstem respiratory control structures, shown in imaging [3, 40] and pathological studies [41] to be significantly damaged in intractable epilepsy patients who succumb to SUDEP. Grey matter volume decline in periaqueductal grey, mesencephalic reticular formation and raphe nuclei in SUDEP and high SUDEP risk patients suggests damage to structures that can influence breathing [42, 43]. Neuropathological examination of SUDEP brainstems [41] shows structural injury in neuropeptidergic and monoaminergic systems, specifically somatostatin neurons and neurokinin 1 receptors, and serotonin transporter activity, in the medullary raphe and rostral ventrolateral medulla, in the region where the pre-Botzinger complex nucleus is located. Specifically, these findings indicate dysfunction of serotonergic and other inputs into respiratory nuclei and provide crucial anatomical and functional substrates that likely mediate respiratory dysfunction seen in near-SUDEP/SUDEP patients.

Serotonergic raphe neurons in the brainstem, which play a major role in chemoreception[9] and arousal[10], have long been suspected to play a role in SUDEP and SIDS[13, 14]. The DBA1 & DBA2 mouse models of seizures and SUDEP, have serotonergic abnormalities and exhibit fatal apnea in the post-ictal state if not resuscitated [16, 44]. Respiratory arrest and death is preventable with SRI pre-treatment [17, 18] most likely due to promotion of respiratory rhythmogenesis and/or arousal rather than increased ventilation [17]. In human seizures, SRIs were found to correlate with reduced severity of hypoxemia [19] a finding confirmed in our study. These SRI effects were previously unexplained, since no reduction in seizure severity was noted [19]. Our findings suggest that ICA is a likely explanation for ictal hypoxemia, and mitigation of hypoxemia is primarily mediated by SRI prevention of ICA occurrence. However, SRIs did not shorten ICA duration, suggesting that this is driven by factors other than serotonergic tone. Bateman et al. [19] did not study ICA incidence but made a similar observation on the absence of SRIs’ impact on ICA duration and hypoxemia duration. In our study, SRIs did not influence electroclinical seizures characteristics, including PGES, which is unsurprising since therapeutic SRI doses have little anticonvulsant effect [19, 45]. SRIs can display proconvulsant properties only at toxic doses [46]. In one study, SRI use in patients with epilepsy neither protected nor elevated seizure-related mortality.[47] However, seizure-related deaths were a broad category, not limited to SUDEP. Further, mood and anxiety disorders (usually treated with SRIs) are associated with worse seizure control, and thus possibly increased mortality risk [48].

GABA, a major respiratory neurotransmitter along with glutamate and glycine, is responsible for tonic inspiratory and expiratory neuronal inhibition, and contributes to respiratory rhythm generation [20, 49] and vagal ventilatory reflexes [50]. BZDs are positive allosteric modulators of GABA receptors [21, 22]. Therapeutic doses, in the absence of other respiratory suppressants, do not significantly reduce respiration, but can do so in healthy subjects in high doses or with rapid intravenous administration[24]. On the contrary, therapeutic dose of BZDs may have stimulant respiratory effects, by enhancing phase switching and increasing respiratory rate; benzodiazepines have been used effectively to prevent central sleep apneas[51]. Animal studies show augmentation in resting respiratory rate[25], due to shortened inspiratory and/or expiratory cycles; in a Rett syndrome mouse model, prone to severe apneic episodes, both BZD and a serotonin agonist showed significant apnea reduction[52].

We found BZDs to be associated with shorter ICA duration and with less profound decreases in SpO₂ and nadirs in the pre-convulsive phase of GCS, without reducing seizure duration. In contrast to SRIs, BZDs significantly mitigated seizure severity through shortened convulsive phases of GCS. BZDs were associated with shorter PGES durations, which may be explained by their effects on shortening the generalized convulsive phase, and so patients on these drugs may recover faster from GCS; PGES is associated with greater post-ictal obtundation and slower recovery.[53] Similar to SRI use, one study showed increase incidence of all-cause mortality in patients with delayed development and nitrazepam usage, mostly due to aspiration pneumonia.[54] On the other hand, recent evidence suggests that SUDEP incidence may be higher in patients not on BZDs [55].

We did not find an additive effect upon ICA in the SRI/BZD group. This may be due to a small number of patients on both drugs, unbalanced groups or the different effect of each drug on ICA, decreasing occurrence vs shortening ICA.

In our study, we found no effect of either class of drug on PCCA, suggesting that PCCA may have a higher threshold for prevention effects than ICA or else, the number of PCCA seizures was too small for an effect to be seen. A recent study with a larger number of seizures (558 in 218 patients) found that longer ICA duration was related to PCCA presence [27] and therefore, preventing ICA may have relevance in the prevention of PCCA.

Similarly, we did not find any effect of either class of drug on hypoxemia after GCS. Unfortunately, we did not have CO₂ measurements, which are a more reliable indicator of respiratory depression after a seizure than hypoxemia, due to the non-linear relationship between partial pressure of oxygen (PaO₂) and SpO₂.

Several other limitations exist in our study. The cohort studied is heterogeneous in several respects. The SRI group contain individuals with depression who potentially have lowered endogenous serotonergic tone although this is difficult to define and quantify. Our study was not sufficiently powered to examine different types of SRIs, BZDs, and their dosages and there was no objective measure of the SRIs or BZDs use by the patients, for example, by measuring serum levels. Thus, a prospective controlled study examining types and dosage of SRIs and BZDs and blood levels will be necessary to confirm our findings. We also did not examine comparative effects of other seizure medications on breathing and electroclinical seizure characteristics, primarily because these do not have any known direct impact on respiration. However, they may influence other electroclinical seizure features and merit further study. Other confounding variables may have influenced our findings although we found no differences between SRI, BZD and control groups in gender, age, epilepsy duration, awake and sleep states, epilepsy type, or oxygen administration and post-ictal intervention. Larger studies with increased number of PCCA and ideally with CO₂ measurements would be crucial to confirm a possible protective effect of SRIs and/or BZDs on respiratory depression after GCS.

CONCLUSIONS

This study suggests that SRIs and BZDs may influence ictal central apnea (ICA) and hypoxemia during seizures, both of which have been implicated in SUDEP pathomechanisms and are potential pharmacological SUDEP interventions. BZDs may have beneficial effects on electroclinical seizure features associated with reduced seizure severity. The importance of this study is that it may form the basis of a prospective randomized study of SRIs and BZDs in patients with epilepsy to determine their possible role in human SUDEP risk-reduction.

Highlights.

• Seizures in patients taking serotonin reuptake inhibitors (SRIs) have lower relative risk for Ictal Central Apnea (ICA) occurrence.

• Seizures in patients on chronic benzodiazepines (BZDs) treatment are associated with shorter ICA duration, but not with ICA incidence.

• Seizures in the BDZ group, have shorter generalized convulsive seizures, as well as shorter post-ictal generalized EEG suppression durations.

• Seizures in patients on SRIs or BZDs are associated with less severe hypoxemia than in the control group (no SRIs or BZDs).