Use of Fluoxetine to Augment the Inter-Ictal Hypercapnic Ventilatory Response in Patients with Epilepsy: A Pilot Study
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.359160
Source of Support: None, Conflict of Interest: None
Keywords: Epilepsy, fluoxetine, HCVR, SSRI, SUDEP
SUDEP is a major cause of premature death in patients with epilepsy. Severe respiratory dysfunction following generalized convulsive seizures (GCSs) often precedes death in SUDEP., Thus, severe peri-ictal respiratory dysfunction has been proposed as a biomarker of SUDEP risk. Other competing proposed mechanisms of SUDEP include cardiac abnormalities and autonomic failure, which are mainly based on the reported association of cardiac and autonomic changes in longstanding drug-resistant epilepsy patients as well as peri-ictal changes that occur with GCS.,, several neurocardiac channelopathies have been identified in some patients with epilepsy which could also support a cardiac hypothesis.,,
Serotonin (5-HT) neurons in the brainstem mediate central chemoreception (CCR) by sensing a rise in PaCO2 via low tissue pH and stimulate respiration and induce arousal to restore homeostasis., This CCR can be quantified by measuring the change in minute ventilation (VE) in response to increasing end-tidal (ET) CO2, also known as the hypercapnic ventilatory response (HCVR), by using a modified hyperoxic CO2 rebreathing technique. An attenuated inter-ictal HCVR correlates with severe respiratory dysfunction related to GCS, identifying low HCVR as a surrogate of SUDEP risk.
Retrospective studies suggest that selective serotonin reuptake inhibitors (SSRI) use is associated with less severe peri-ictal hypoxemia., The mechanisms for this are unclear but may involve augmentation of the HCVR. SSRIs, including fluoxetine, have been studied in animals and humans to modify the HCVR., Fluoxetine was associated with reduction of the HCVR in patients with panic disorder (PD), but there are reasons to believe this response may be different in patients with epilepsy without PD, a group that has not been studied., Therefore, we conducted a randomized clinical trial to assess the feasibility of such a study in patients with epilepsy.
This was a randomized controlled trial to assess the feasibility of studying the effect of fluoxetine on HCVR in patients with epilepsy. The study was approved on 10/14/2016 by the University of Iowa Institutional Review Board (ClinicalTrials.gov registration number NCT029296670).
We recruited consecutive patients with epilepsy, aged 18–75 years, at the University of Iowa Hospitals and Clinics from February 2017 to February 2019. Written informed consent was obtained for all subjects. Eligible subjects underwent inter-ictal HCVR measurement; HCVR slope was calculated for each subject using a previously described method. [Table 1] describes inclusion and exclusion criteria for randomization.
Procedures and interventions
We used a screening questionnaire and electronic medical records to identify covariates such as demographics, medical history, and active medications, including anti-epileptic drugs.
After initial enrollment, Patient Health Questionnaire (PH-Q)-9 was assessed as well as the measurement of HCVR to determine further eligibility for randomization to study the drug. Eligible subjects were randomized equally between fluoxetine and placebo [Figure 1] to a 6-week titration plan: week 1-10 mg/day, week 2-20 mg/day, weeks 3 and 4-40 mg/day, week 5-20 mg/day, week 6-10 mg/day, and then stop from week 7 onward. The HCVR and PHQ-9 were repeated at the end of week 4. Follow-up telephone calls were made during weeks 1, 3, and 7 for assessment of compliance and adverse effects. Two subjects proceeded with previously arranged admissions to the epilepsy monitoring unit (EMU) for video EEG with comprehensive cardiorespiratory monitoring as previously described.
Primary outcomes were recruitment and retention rates. Secondary outcomes included a) change in PHQ-9 score, b) change in baseline and maximum VE, c) change in HCVR slope, d) difference in duration of abnormal breathing (bradypnea and/or apnea) in the peri-ictal period, e) difference in oxygen saturation during the peri-ictal period, and f) difference in transcutaneous CO2 level in the peri-ictal period.
No sample size calculation was performed.
Randomization and Blinding.
Using block randomization with a block size of 4, the randomization sequence was generated using a tool available at www.randomization.com. The subjects, investigators, care team, and statistician were all blinded to treatment allocation.
All analyses utilized the generalized linear mixed modeling framework, specifying an identity link or a log link for normally distributed or right-skewed outcomes, respectively. A random effect was included for study ID. For normally distributed outcomes, the mean difference, standard deviation, 95% confidence interval, and P between groups were calculated (right-skewed outcomes used mean ratio). We set the significance cutoff at α = 0.05.
Participant flow, recruitment, follow-up, losses, and exclusions
[Figure 1] depicts the participant flow in the study. Of the 126 potentially eligible subjects approached, 30 were enrolled and underwent further screening procedures before randomization to an intervention. Twenty-two eligible subjects were randomized equally (1:1) to one of the treatment arms. Demographics and characteristics of the randomized participants are presented in [Table 2].
One subject in the placebo group lost for the final telephone follow-up. All remaining participants completed the protocol. Secondary outcomes for the subject lost for the final telephone call follow-up were available; therefore, these were included for analyses.
Compliance and adverse effects of the intervention
Compliance was excellent in both groups. No serious adverse events were recorded except for one subject who reported excessive sleepiness, feeling tired and lethargic, and a feeling of a swollen tongue not visible on physical examination. This subject inadvertently deviated from the scheduled titration protocol, which was deemed to be the likely cause. Once back on schedule, symptoms resolved.
Thirty subjects were enrolled, with 22 randomized to an intervention, one of whom was lost for the final telephone follow-up. Therefore, the mean recruitment rate of the study was 3.8 subjects every 3 months with a retention rate of 100% in fluoxetine and 95% in placebo groups, respectively.
There was no significant difference between the two groups for PHQ-9 scores (mean ratio: 1.52, 95% CI: 0.36–6.08, P = 0.55), HCVR slope (mean difference: 0.12, 95% CI: 0.87–0.62 L/min/mm Hg, P = 0.75), baseline VE (mean ratio: 0.95, 95% CI: 0.75–1.20, P = 0.65), or maximum VE during HCVR measurement (mean ratio: 0.96, 95% CI: 0.76–1.2, P = 0.71). The two subjects who underwent video EEG study with comprehensive respiratory monitoring did not have any recorded seizures. Therefore, the effect of the intervention on peri-ictal respiratory abnormalities could not be evaluated.
There is an urgent need to develop a therapy to significantly reduce SUDEP risk. Our previous study demonstrated that a low inter-ictal HCVR slope correlates with greater severity of peri-ictal respiratory dysfunction after GCS, suggesting that it is a potential biomarker for SUDEP risk., Given the important role of 5-HT in CCR, we conducted a randomized clinical trial to assess the feasibility of a study using fluoxetine in patients with epilepsy to modify their HCVR. Our successful conclusion of the trial demonstrates such feasibility despite stringent subject selection.
SSRI agents usually take several weeks to exert a positive effect on mood., We assumed that a similar dose and duration of therapy may be sufficient to modify the HCVR. Therefore, we planned a 6-week titration schedule which was well tolerated.
We did not find any statistically significant differences in secondary outcomes between the two groups: HCVR slope, baseline VE, or maximum VE. These findings are not totally unexpected given the study was not powered to measure small differences between the two groups. Biological variables, particularly age, gender, and body size are known to influence HCVR, but it remains unclear if these variables also influence fluoxetine's ability to alter HCVR in any way. If they do, they could have affected the results as we were not able to control them in this study. This would be particularly true for age as this variable was significantly different between the two groups. Alternatively, the dose and duration of fluoxetine to alter the HCVR may be different than that was used in the study. Thus, the results of our study do not answer the question of whether fluoxetine increases the HCVR slope in patients with epilepsy and, if so, what the optimal dose and duration would be.
Interestingly, previous observational studies in patients with epilepsy have suggested that use of SSRI agents was associated with less severe peri-ictal hypoxemia., We were not able to evaluate the effect of fluoxetine on peri-ictal O2 saturation or CO2 changes as the two subjects who underwent VEEG with comprehensive cardiorespiratory monitoring did not have any seizures during the study. Our results suggest that with stringent inclusion/exclusion criteria, it may be difficult to enroll adequate numbers of patients prior to EMU admission to study the effect of SSRIs on HCVR and peri-ictal respiratory changes.
This phase II study demonstrates the feasibility of a prospective clinical trial in patients with epilepsy to evaluate whether fluoxetine modifies the HCVR slope. Future studies that utilize different doses and/or durations of treatment may be necessary to ascertain the effect on HCVR slope in patients with epilepsy.
Financial support and sponsorship
This study was supported by Citizens United for Research in Epilepsy (CURE) (award number: 411673), and the National Institute of Health CTSA (program grant: UL1TR002537).
Conflicts of interest
There are no conflicts of interest.
[Table 1], [Table 2]