The effect of obstructive sleep apnea severity on PTSD symptoms during the course of esketamine treatment: a retrospective clinical study

Abstract

Study Objectives:

Intranasal administration of esketamine is Food and Drug Administration–approved for treatment-resistant depression. In a recent retrospective case series, we show that it has promise in reducing symptoms of posttraumatic stress disorder (PTSD) as well. Untreated obstructive sleep apnea (OSA) is prevalent among veterans with PTSD and has been shown to interfere with other PTSD treatments. In the current study, we examined whether OSA impacts esketamine’s effectiveness in reducing symptoms of PTSD or depression.

Methods:

Participants were 60 veterans with a diagnosis of major depressive disorder and PTSD who received intranasal esketamine treatment at the San Diego Veterans Affairs (VA) Medical Center. We used growth-curve modeling to examine changes in depression and PTSD symptoms following esketamine treatments and, in the subset of individuals screened for OSA (n = 24, all prescribed positive airway pressure therapy), examined the impacts of OSA severity on these trajectories.

Results:

We first showed that both PTSD and depressive symptoms significantly decreased over the course of esketamine treatment. In the subset of veterans screened for OSA, individuals with lower OSA severity reported the greatest reduction in PTSD symptoms, while veterans with the most severe OSA reported the least reduction in PTSD symptoms. Depression response was not affected by severity of OSA in this analysis.

Conclusions:

Veterans with PTSD and depression tend to benefit from esketamine treatment, but OSA may interfere with esketamine effectiveness. Comorbid OSA should be assessed for and treated to maximize esketamine’s benefits in PTSD.

Citation:

Titone MK, Hunt C, Bismark A, et al. The effect of obstructive sleep apnea severity on PTSD symptoms during the course of esketamine treatment: a retrospective clinical study. J Clin Sleep Med. 2023;19(12):2043–2051.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Intranasal esketamine shows promise for treating symptoms of posttraumatic stress disorder (PTSD) and depression in veterans; however, untreated obstructive sleep apnea is prevalent among veterans with PTSD and has been shown to interfere with other PTSD treatments. We conducted this study in order to explore if obstructive sleep apnea impacts esketamine’s effectiveness in reducing symptoms of PTSD or depression.

Study Impact: Results showed that veterans with the highest obstructive sleep apnea severity reported the least reduction in PTSD symptoms following esketamine treatment. This provides evidence that severe obstructive sleep apnea may interfere with the efficacy of intranasal esketamine treatment for PTSD.

INTRODUCTION

Posttraumatic stress disorder (PTSD) is highly prevalent in the veteran population, with estimates ranging from 12% to 25%, twice the 6.8% lifetime prevalence in the US general population.^1–3 PTSD is debilitating for many veterans; US Vietnam and Operation Iraqi Freedom/Operation Enduring Freedom veterans with PTSD demonstrate increased functional disability, decreased quality of life, and decreased objective living conditions when compared with veterans in the same cohorts without PTSD.⁴ PTSD is also associated with increased suicide attempts, poorer work and family functioning, and negative consequences on intimate relationships, evincing high societal costs.^5,6

Current treatments for PTSD comprise 2 main categories: psychotherapy and pharmacotherapy. Psychotherapy in the form of cognitive behavioral therapy (eg, cognitive processing therapy) or exposure therapy (eg, prolonged exposure) is the first-line treatment for PTSD and is efficacious for roughly half of veteran and civilian samples; however, early dropout rates are high for these psychotherapies, especially in veterans, suggesting low tolerability.^7,8 Pharmacotherapy treatment options for PTSD are more limited, but selective serotonin reuptake inhibitors (SSRIs), the first-line pharmacological treatment, show small effect sizes for a minority of treatment-seeking patients.^9–11 Given the limitations of the current approaches, many individuals who undergo treatment do not fully respond (eg, they fail to achieve a PTSD symptom severity score on the Clinician Administered PTSD Scale [CAPS-5] of < 20 after 8 to 12 wk of adequate antidepressant dosing and psychotherapy attendance); these patients may be considered to have treatment-resistant PTSD.^12,13

Ketamine, a glutamate receptor antagonist, has shown promise in reducing symptoms of depression.^14–16 An intranasal formulation of (S)-ketamine (esketamine) has been approved for the treatment of treatment-resistant depression by the Food and Drug Administration (FDA). Clinical trials have found that intranasal administration of esketamine also has rapid antidepressive effects,^17,18 although meta-analyses suggest that effects may be lower than those observed with intravenous racemic ketamine.¹⁹ Esketamine, the FDA-approved formulation of ketamine, has been utilized for treatment-resistant depression in veterans at several Veterans Affairs (VA) hospitals, including at the San Diego VA Medical Center. A recent retrospective case-series analysis of 35 veterans with depression and PTSD treated with esketamine from this clinic demonstrated a clinically meaningful reduction in PTSD symptoms in ∼50% of veterans in the study over the course of 4 weeks.²⁰ These results, interestingly, correspond with some other open-label and clinical trials, showing that racemic intravenous ketamine has rapid-acting effects on reducing PTSD symptoms^14,21,22 in civilians. However, a recent large multisite, double-blind, randomized controlled trial (RCT) of racemic intravenous ketamine for PTSD in veterans and active-duty military personnel showed no differences between ketamine and placebo in ameliorating PTSD symptoms.²³

Thus, an outstanding question in the field is whether there is a difference between civilian and military PTSD that might explain why early ketamine trials in civilian PTSD appear promising but an RCT in the veteran population did not demonstrate superiority of ketamine over placebo. One reasonable hypothesis is that veterans may have certain common characteristics (either in the type of PTSD, or in comorbid medical or psychiatric illnesses) that lower the efficacy of ketamine in ameliorating PTSD symptoms. Given the nascent research on esketamine, barriers or facilitators of esketamine effectiveness for PTSD are unknown. Obstructive sleep apnea (OSA) may be one such factor.

OSA’s hallmark features are repetitive collapse of the pharyngeal airway, causing hypoxemia, hypercapnia, swings in intrathoracic pressure, increased sympathetic tone, and sleep fragmentation, with associated daytime symptoms.^24,25 The apnea-hypopnea index (AHI; number of apneas plus hypopneas per hour of sleep) is the most commonly used metric of OSA severity, with mild OSA starting at AHI ≥ 5 events/h.²⁶ OSA is known to cause sleep fragmentation, excessive daytime sleepiness, and worsening sleep quality for those who are affected by the disorder.²⁷ General-population studies of OSA in veterans (ie, not sleep disorder–specific referrals) suggest lifetime prevalence rates between 27% and 44%.²⁸ Prevalence rates are even higher among veterans with PTSD, with an estimated 67–83% of veterans with PTSD having co-occurring OSA.^29–32 Moreover, OSA is the most common respiratory service-connected condition, and rates are expected to rise with increasing age and obesity within the veteran and general population.^33,34

Untreated OSA seems to interfere with successful PTSD treatment.³⁵ Two studies found that those with OSA showed markedly less PTSD symptom improvement than those without OSA in evidence-based, trauma-focused PTSD treatment.^36,37 The relationship between OSA and PTSD is bidirectional, where PTSD is thought to worsen OSA severity measurements through proclivity to arousal, and the impact of OSA on sleep has been shown to impact PTSD outcomes.^38–40 How individual OSA endotypic traits and PTSD phenotypic clusters interact is still being established.⁴¹ Importantly, those with OSA who were treated with positive airway pressure (PAP; the gold-standard treatment for OSA) showed significantly more improvement in response to psychotherapy than those who were not engaging in PAP treatment.^36,37 OSA treatment with PAP, mandibular advancement device, and hypoglossal nerve stimulation have shown daytime symptomatic benefits for individuals with PTSD.^42–44 Unfortunately, OSA screening and treatment are often not a part of clinical treatment protocols for mental health patients, as the symptoms of disrupted sleep are often assumed to be associated with the “primary” disorder (eg, PTSD).

Given the evidence reviewed above, it is clear that the co-occurrence of PTSD and OSA has negative additive effects that worsen disease burden and interfere with the efficacy of PTSD treatment.^45,46 OSA interferes with both learning and sleep-dependent memory consolidation via sleep fragmentation and hypoxemia,^47–52 and these memory deficits may be reversible with PAP therapy.⁵³ It is possible that the presence of OSA may interfere with ketamine’s therapeutic action in PTSD, given that ketamine affects N-methyl-d-aspartate (NMDA) receptors and glutamatergic transmission, which also play a role in increasing synaptic plasticity, learning, and memory reconsolidation.⁵⁴ NMDA receptors have also been implicated in fear conditioning, and it is possible that one of ketamine’s therapeutic mechanisms is via the facilitation of fear extinction in PTSD.⁵⁵ However, if sleep quality is worse due to the presence of OSA, then learning, memory reconsolidation, and increases in synaptic plasticity may not occur optimally, and ketamine treatment may have a reduced ability to treat PTSD symptoms. Additionally, higher OSA severity increases sympathetic nervous system activation (ie, “fight or flight” responding), which may interfere with ketamine’s ability to facilitate fear extinction in PTSD.

To add to the nascent literature on esketamine treatment, we first aimed to examine the effects of esketamine treatment on PTSD symptoms over time. There are limited data on the efficacy of esketamine (as opposed to racemic intravenous ketamine) in ameliorating PTSD symptoms, so better characterization of the trajectories of response is helpful. To better understand this understudied pharmacotherapy and its relationships to sleep and PTSD outcomes, our second aim was to examine whether OSA interfered with the effects of intranasal esketamine treatment on PTSD symptoms among veterans. We hypothesized that veterans with more severe OSA would see attenuated improvement in PTSD symptoms following acute intranasal esketamine treatment compared with veterans with less severe or treated OSA.

METHODS

Data were collected from 60 consecutive veterans who were outpatients at the San Diego VA Neuromodulation Program between the dates of January 2020 and April 2021. Veterans were included if they were undergoing esketamine treatment in the VA San Diego Neuromodulation Program and had valid PTSD Checklist for Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (PCL-5), questionnaire data collected before and/or during esketamine treatment. This study was approved as an institutional review board (IRB) exemption by the local VA IRB (IRB 1223219). The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.

Referrals to the Neuromodulation Program came from the veterans’ primary VA psychiatrist, requesting evaluation for treatment-resistant depression. During initial consultation with a physician, the veteran and physician collaboratively decided which of 3 treatments offered by the Neuromodulation Program would be most appropriate: repetitive transcranial magnetic stimulation, esketamine, or electroconvulsive therapy. Participants were invited to start esketamine treatment if they (1) showed no or insufficient response to at least 2 traditional antidepressants in the past and (2) had a 9-item Patient Health Questionnaire (PHQ-9) score of > 15 at that baseline consultation with the physician. Participants were excluded from esketamine treatment if they had known serious medical contraindications (eg, history of seizures, severe hepatic disease, cystitis), psychosis, past or present abuse of ketamine, or neurocognitive disorder.

Participants were included in our analysis if they underwent esketamine treatment and had valid PCL-5 scores collected before and/or during treatment. Sixty participants completed the PCL-5, 24 (40%) of whom were screened for OSA by a primary care provider or psychiatrist in the sleep medicine clinic and had completed sleep studies.

Measures

Clinical measures

All clinical measures were gathered as a part of routine clinical care in the esketamine clinic. Veterans completed the PHQ-9 and the PCL-5 prior to each esketamine treatment within the clinic.

PCL-5: This 20-item self-report measure assesses PTSD symptoms, and each item directly reflects the Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-5), diagnostic criteria for PTSD.⁵⁶ Participants indicate on a 0 (never) to 5 (extremely) Likert scale as to how much they were bothered by that symptom within the past week. Higher scores indicate more severe PTSD symptom severity. This measure has excellent psychometric properties.^57–59

PHQ-9: The PHQ-9 is a 9-item questionnaire that assesses depression symptoms,⁶⁰ with each question directly mapping onto the DSM-5 diagnostic criteria for major depressive disorder. Participants respond on a 0 (not at all) to 5 (nearly every day) Likert scale as to how often they were bothered by each depression symptom in the past 2 weeks, with higher scores indicating more severe depression. The PHQ-9 has shown a strong correlation with other well-established measures of depression, along with good sensitivity and specificity.⁶¹

Chart review

Data from chart review were accessible via a Computerized Patient Record System, the electronic medical record system utilized by VA. A chart review of the past 5 years was performed to gather demographic (eg age at visit, sex, race/ethnicity) and other data. We reviewed the diagnostic codes that had been entered into the medical record system, combined with the physician’s impression from the first session, to gather past mental health diagnostic history, including the following: history of ketamine/esketamine treatment, history of electroconvulsive therapy, lifetime PTSD, lifetime primary psychotic disorder, lifetime anxiety disorders, lifetime alcohol-use disorder, chronic pain, years in mental health treatment, past number of suicide attempts, past number of psychiatric hospitalizations, and past total number of antidepressant trials.

Esketamine treatment

Esketamine was administered intranasally and occurred as part of routine clinical care in the Neuromodulation Program for treatment-resistant depression. Esketamine treatment generally occurred twice per week for the first 4 weeks of treatment (for a total of 8 treatments). Veterans who chose to continue esketamine after the first treatments were transitioned to a once-per-week-or-less dosing schedule. Esketamine was always started at 56 mg at the first treatment and was titrated up to 84 mg (high dose) or down to 28 mg (low dose) depending on clinical response. By the eighth treatment, the majority of veterans in the study were titrated up to the high dose. Although treatment could continue as long as it was clinically warranted, we chose to restrict our study to the first 13 weeks of treatment (ie, 17 doses) since (1) this number of time points provided sufficient power to conduct our main analyses, (2) the proportion of participants with missing data exceeded 75% following this time point and could lead to biased estimates, and (3) 13 weeks of treatment corresponded to 3 months, which is a common endpoint of clinical trials including those that have investigated esketamine as a depression treatment.⁶²

OSA home sleep tests

Sleep apnea metrices were obtained predominantly through Type III home sleep testing within our center as clinically indicated (Nox Medical, Suwanee, GA, USA). Some individuals had in-laboratory polysomnography performed at outside facilities prior to entering our care. Apneas and hypopneas per hour (AHI) were scored based on desaturations with home sleep testing using American Academy of Sleep Medicine 1A criteria.²⁶ Each of these sleep studies were conducted due to a clinical suspicion of OSA. With regard to treatment, all veterans with confirmed OSA diagnoses in our analyses were on auto-PAP devices.⁶³ Because continuous positive airway pressure (CPAP) adherence data were only available for n = 15 participants, they were not examined in our main analyses; however, we do report average CPAP usage per night in the last 90 days and its correlation with the AHI for descriptive purposes.

Statistical analysis plan

Multiple measurements per individual occurred over the course of the treatment; thus, we utilized growth curve modeling, which facilitated an examination of change trajectories over time. Analyses were conducted using SPSS version 28 (IBM Corporation, Armonk, NY).⁶⁴

Bivariate Pearson correlations between variables of interest were conducted prior to primary analyses. Independent-samples t tests were conducted to test if there were significant differences in demographic/baseline characteristics between the OSA and non-OSA data groups.

In preparation for the primary analyses testing the effect of OSA severity during esketamine treatment on PCL-5 and PHQ-9 scores, we tested 3 basic candidate models for each outcome (PCL-5 and PHQ-9) to find the best model fit. We tested (1) a random intercept model, (2) a random intercept and fixed slope model, and (3) a random intercept and random slope model. For the PCL-5 outcome model, the random intercept and fixed slope model provided the best fit. For the PHQ-9 outcome model, the random intercept and random slope model was the best fit.

Within the best-fitting model, we conducted 2 separate sets of our analyses. The first set of analyses focused on testing the effect of esketamine dose on PTSD symptoms (ie, PCL-5 scores) and depression symptoms (ie, PHQ-9 scores). Here, we hypothesized a negative effect of time (ie, number of esketamine doses, ranging from 1 to 17), such that both PCL-5 and PHQ-9 scores decreased with more esketamine doses.

The second set of analyses focused on testing whether treatment response (ie, PHQ-9 and PCL-5 decrease) was moderated by OSA severity, which was operationalized as the AHI score prior to treatment. Specifically, we tested the effect of (1) time, (2) AHI score, and (3) time × AHI interaction, which tested the hypothesis that the degree of symptom decrease across sessions was moderated by OSA severity. We hypothesized a negative time × AHI interaction, such that the slope of symptom decrease would be less steep for participants with higher AHI scores. Significant 2-way interactions were followed up with simple slopes evaluating the relationship between time (ie, esketamine dose) and clinical outcome (ie, PCL-5, PHQ-9) at a low AHI (–1 SD), moderate AHI (mean), and high AHI (+1 SD), which, in this sample, corresponded to an AHI of 0, 15, and 30 events/h, respectively. In addition, we used these simple slopes to approximate the clinical significance of the esketamine treatment for these 3 AHI levels by estimating the degree of symptom decrease that would theoretically occur by session 17. Since only n = 24 participants had AHI data available, this second set of analyses focused on OSA were only conducted on these 24 participants.

RESULTS

Analyses were conducted on 60 participants (mean age = 47.58, SD = 11.93; 28.3% female, 68.3% White, 20% Hispanic, 8.3% Black, and 3.3% Native Hawaiian or other Pacific Islander; see Table 1 for all descriptive statistics). We conducted independent-samples t tests to test if there were significant differences in demographic/baseline characteristics between the OSA and non-OSA data groups; the only variable that differed between groups was the mean number of esketamine sessions completed, with the non-OSA data group completing more treatment sessions than the OSA data group: t(60) = 2.07, P = .043. Among those with OSA data (n = 24), the mean AHI was 14.8 (SD = 14.5) events/h, while mean CPAP usage was 5.43 hours for the last 90 days (SD = 2.97 hours).

Table 1.

Descriptive statistics for the total sample and in participants with and without OSA data.

Participant Characteristics	Total Sample (n = 60)	With OSA Data (n = 24)	Without OSA Data (n = 36)
Age, y	47.6 (11.9)	48.3 (12.2)	47.1 (11.9)
Race
White	41 (68.3%)	17 (70.8%)	24 (66.7%)
Black	5 (8.3%)	3 (12.5%)	2 (5.5%)
Hispanic	12 (20%)	3 (12.5%)	9 (25%)
Other	2 (3.3%)	1 (4.2%)	1 (2.8%)
Sex
Female	17 (28.3%)	5 (20.8%)	12 (33.3%)
Male	43 (71.6%)	19 (79.2%)	24 (66.6%)
Number of psychiatric hospitalizations	2.2 (3.4)	2.8 (4.2)	1.6 (2.7)
Number of suicide attempts	1.1 (1.6)	1.3 (1.9)	.92 (1.4)
Total number of antidepressant trials	3.6 (1.8)	4.0 (1.9)	3.3 (1.7)
Alcohol use disorder (lifetime)	26 (43.3%)	12 (50.0%)	14 (38.8%)
Primary psychotic disorder (lifetime)	0 (0%)	0 (0%)	0 (0%)
Other anxiety disorder (lifetime)	29 (48.3%)	11 (45.8%)	18 (50%)
Chronic pain	42 (70%)	15 (62.5%)	27 (75%)
PTSD (lifetime)	55 (91.6%)	24 (100%)	31 (86.1%)
History of electroconvulsive therapy	16 (26.7%)	3 (12.5%)	13 (36.1%)
History of ketamine/esketamine	17 (28.3%)	8 (33.3%)	9 (25%)
Years in mental health treatment	16.4 (10.6)	13.9 (8.3)	18.1 (11.8)
Mean number of esketamine treatments	10.9 (6.1)	9.0 (6.2)	12.2 (5.8)
PHQ-9 score (pretreatment)	19.2 (4.3)	19.3 (5.2)	18.7 (4.4)
PCL-5 score (pretreatment)	50.3 (15.6)	50.7 (15.9)	48.6 (14.6)
AHI	N/A	14.8 (14.5)	N/A

Data are presented as mean (SD) or n (%). AHI = apnea-hypopnea index, OSA = obstructive sleep apnea, PCL-5 = PTSD Checklist for Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, PHQ-9 = 9-item Patient Health Questionnaire.

Table 2 presents the bivariate correlations between variables of interest (baseline PHQ-9 score, baseline PCL-5 score, and AHI [OSA data group only]). Bivariate correlations and independent-samples t tests were conducted between the outcome variables of interest (PHQ-9 score, PCL-5 score) and demographic variables (sex, race/ethnicity, and age). The only demographic variable significantly related to an outcome variable was age, which was negatively correlated with PCL-5 score (r = –.267). Age was thus included as a covariate in the subsequent main analyses involving PCL-5 score as an outcome. Notably, there was no correlation between AHI and CPAP usage (r = –.10, P = .774), indicating that adherence was similar across veterans with varying AHI levels.

Table 2.

Bivariate correlations between variables of interest.

Measure	PHQ-9	PCL-5	AHI
PHQ-9 (baseline)	–	.427*	.075
PCL-5 (baseline)		–	−.131
AHI			–

^*P < .05. AHI = apnea-hypopnea index, PCL-5 = PTSD Checklist for Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, PHQ-9 = 9-item Patient Health Questionnaire-9.

Main analyses

Our first analysis examined the main effect of treatment session number on PTSD symptoms and depressive symptoms. Our results showed that PTSD symptoms (PCL-5 scores) were negatively associated with treatment session number (b = –0.95, t[474.29] = –10.40, P < .001), meaning that PTSD symptoms tended to decrease over time in treatment. Our results showed a similar relationship with depressive symptoms (PHQ-9 scores; b = –0.22, t[46.00] = –4.60, P < .001), which decreased over the course of treatment.

Our second set of analyses tested the interaction between AHI levels and time (treatment session number) on (1) PCL-5 score and (2) PHQ-9 score. Results indicated that there was a significant interaction between AHI and treatment session number predicting PCL-5 score: b = 0.40, t(182.54) = 2.45, P = .015. As shown in Figure 1, this interaction appeared to be driven by there being a less steep reduction in PCL-5 scores with higher AHI levels, such that the per-session decrease on the PCL-5 was highest at an AHI level of 0 events/h (b = –1.72, t[173] = –7.11, P < .001), somewhat lower for an AHI level of 15 events/h (b = –1.32 t[173] = –8.00, P < .001), and lowest for an AHI of 30 events/h (b = –0.91, t[173] = –4.18, P < .001).

Figure 1. Interaction between AHI score and treatment number on PCL-5 scores (model-derived values representing the interpolated slope from the predicted PCL-5 score at final session).

AHI = apnea-hypopnea index, PCL-5 = PTSD Checklist for Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition.

In contrast, there was no significant interaction between esketamine dose and AHI in predicting PHQ-9 scores: b = 0.06, t(20.09) = 0.69, P = .496. Thus, the ability of esketamine to reduce depression symptoms did not appear to be affected by OSA severity, as shown in Figure 2.

Figure 2. Nonsignificant interaction between AHI score and treatment number on PHQ-9 scores (model-derived values representing the interpolated slope from the predicted PHQ-9 score at final session).

AHI = apnea-hypopnea index, PHQ-9 = 9-item Patient Health Questionnaire.

DISCUSSION

We examined how OSA severity affects esketamine treatment on changes in PTSD and depression self-report rating scales. Our results indicated that both PTSD and depressive symptoms significantly decreased over the course of esketamine treatment in veterans. As hypothesized, OSA severity was associated with a significant reduction in the magnitude of response for PTSD symptoms but not for depression. Thus, veterans with the most severe OSA saw the least reduction in their PTSD symptoms over the course of esketamine treatment. Importantly, there was no effect of OSA severity on the reduction in depression symptoms. Thus, our findings suggest that the antidepressant and anti-PTSD effects of esketamine are different, with OSA primarily affecting the latter.

Our findings showing a beneficial effect of esketamine treatment on depressive symptoms replicate similar results in RCTs of esketamine for depression.^17,18 No randomized clinical trials have yet studied the effects of esketamine treatment on PTSD, and these results build upon our previously published dataset on this topic, which includes many of the same participants.²⁰ Our study, although not an RCT, is the largest study to date examining esketamine’s effect on PTSD symptoms; combined with existing evidence, our study provides a preliminary basis for esketamine’s effectiveness and strengthens the rationale for conducting larger RCTs of esketamine for PTSD.

Several civilian studies of intravenous racemic ketamine have shown beneficial effects on PTSD symptoms,^14,22 which are consistent with our results described above with esketamine. Importantly, however, a large multisite RCT was recently conducted within VA comparing a series of intravenous racemic ketamine at 2 different doses with a saline placebo. This trial showed no difference in the reduction in PTSD symptoms between ketamine treatments compared with a saline placebo.²³ Our data suggest that this lack of differential response in veterans with PTSD symptoms may be partially explained by the severity of undiagnosed and untreated OSA in a veteran cohort compared with the general population.⁶⁵ We believe that our data support a new trial of ketamine in veterans.

Our study is the first to our knowledge to find that higher OSA severity showed a lower effectiveness of esketamine for treating PTSD symptoms. Moreover, OSA severity in these same participants was not a significant moderator of the change in PHQ-9 (depression) scores following ketamine treatments. This suggests that ketamine’s improvement of depression vs PTSD symptoms may rely on distinct mechanisms of action. The mechanism of action of ketamine has not been fully elucidated. Ketamine clearly results in rapid changes in circuit functioning^66,67 that may mediate the rapid antidepressant effects, while the sustained/longer-term antidepressant effects of ketamine may involve activation of intracellular pathways involved in synaptic plasticity, including CamKII, BDNF, and MeCP2,⁶⁸ resulting in synaptogenesis of spines within the prefrontal cortex.⁶⁶

Our data (including both impacts of OSA and overall trajectory of symptom response) suggest that the effects on PTSD symptoms, unlike effects observed in depression, take longer to manifest. Thus, 1 possibility is that ketamine helps facilitate fear extinction or minimize fear generalization. Ketamine may facilitate fear extinction learning and/or fear responding in PTSD,^69–71 a process that is thought to be involved in other forms of PTSD treatment, such as prolonged exposure therapy. Several other studies have demonstrated that OSA severity interferes with PTSD treatment effectiveness more generally.^35–37 Specifically, untreated OSA has been shown to reduce the effectiveness of cognitive processing therapy and prolonged exposure therapy for PTSD.^35,36 Thus, our study suggests that esketamine and trauma-focused psychotherapy may have shared mechanisms of action, and the presence of OSA interferes with both treatments in a similar way, perhaps by impairing learning processes involved in fear extinction. Prolonged exposure therapy is known to facilitate fear-inhibitory learning, 1 mechanism by which PTSD symptoms may lessen over the course of treatment.⁷² Evidence suggests that OSA also may interfere with fear inhibition and fear extinction⁷³ through numerous mechanisms and neurobiological pathways. Improved sleep, decrease in repetitive nighttime arousals, and/or increased sustained and restorative sleep may enhance learning and cognitive functioning.⁷³ A final possibility is that ketamine improvements in sleep architecture (eg, decreasing fragmentation, increased slow-wave sleep or more consolidated sleep) may facilitate a long-term reduction in PTSD symptoms.⁷⁴ In individuals with OSA, these improvements in sleep architecture may not manifest as clearly, thus impairing treatment.

Our study exhibited several important strengths, including a unique, well-characterized sample. The integrated nature of VA clinical care allowed us to collect data on OSA from a separate sleep disorders treatment clinic and integrate them with data from the large neuromodulation clinic. Examining these data jointly allowed for a unique analysis of an important but understudied comorbidity (OSA/PTSD) in the context of treatment with a novel pharmacotherapy. We also collected frequent, repeated measures of clinical outcomes, allowing for a detailed examination of treatment effect over time. Given that our measures were collected during the course of normal clinical care in a VA setting, our results have high external validity, allowing for broader generalization of findings to VA clinics.

Our study has several limitations. First, we had a small sample size, limiting the ability to draw strong conclusions from our interaction analysis. As such, our finding should be interpreted with caution and future studies are necessary for confirming our findings, as well as examining moderators of PTSD treatment outcomes. For example, we are unable to examine the mechanisms through which ketamine may affect upper airway dilator dysfunction.⁷⁵ Second, there was limited objective screening for OSA in our sample. The non-OSA data group was not screened for OSA, thus introducing the potential for undiagnosed and/or untreated OSA in that group. Notably, the group not screened for OSA improved in PCL-5 score in a similar amount to the average AHI in the OSA data group (12.92 points and 15.98 points, respectively). One way to interpret this finding is that the group not screened for OSA may have included some participants with undiagnosed OSA, thus having a poorer response to ketamine than the participants with a low AHI in the OSA data group. Third, the time between OSA diagnosis and esketamine treatment initiation was variable. As such, the number of months between when OSA was diagnosed and esketamine treatment initiation was highly variable. This confound could allow for AHI to fluctuate between diagnosis and ketamine initiation.^76,77 Fourth, we were unable to address OSA treatment as a moderator of treatment outcomes. Although we report data pertaining to OSA treatment in the group that was screened for OSA, including basic descriptive statistics for PAP usage and adherence, our small sample size did not allow for extensive analyses on this particular outcome measure. However, the correlation between AHI and PAP usage (r = –.10) was nonsignificant, suggesting that PAP usage is likely not a confounder accounting for the differences in treatment response by OSA severity. Veterans with more severe OSA did not have better or worse PAP usage, so our reported results are likely unrelated to PAP usage. Future research should focus on distinguishing between treated and untreated OSA, examining how OSA treatment with the gold-standard therapy (PAP) affects esketamine treatment effectiveness for PTSD and depression.

Our findings, although preliminary, have important implications for the treatment of PTSD symptoms when comorbid OSA is present, which occurs in a large portion of veterans with PTSD. If replicated, our findings might suggest that OSA should be assessed and screened for in all participants undergoing esketamine treatment who have significant PTSD symptoms because the presence of OSA appears to interfere with esketamine’s efficacy. There is increasing evidence that veterans with PTSD exhibit nontraditional OSA (ie, they are younger, lower body mass index, lower blood pressure).^29,32 As such, the paper-and-pencil questionnaires are limited, and more objective OSA testing should be done prior to starting esketamine treatment.⁷⁸

Additionally, individuals with PTSD have been shown to be more likely to have a low respiratory arousal-threshold (ArTh), meaning that small disruptions in ventilation may facilitate sleep fragmentation.⁷⁹ Veterans with a low ArTh have worse CPAP adherence.²⁵ Thus, veterans with PTSD and OSA are likely vulnerable to sleep fragmentation. It is not known how or if esketamine influences the ArTh. Moreover, esketamine is a ventilatory stimulant and may positively affect upper airway dilator dysfunction,⁷⁵ but it is unclear how PTSD, OSA, and esketamine interact at the level of the ventilatory control system and the upper airway.

Furthermore, our findings suggest the need for future studies that will examine the order and subsequent effectiveness of PAP treatment for OSA and esketamine for PTSD. For instance, treatment of OSA with PAP prior to initiation of esketamine treatment may improve sleep quality, cognitive function, and/or fear learning, thus optimizing the effects of esketamine on PTSD via similar mechanisms.

DISCLOSURE STATEMENT

All authors have seen and approved the manuscript. Funding for this work was made possible for Madison Titone, PhD, by the Department of Veterans Affairs Office of Academic Affiliations Advanced Fellowship Program in Mental Illness Research and Treatment, the Medical Research Service of the San Diego Veterans Affairs Health Care System, and the Department of Veterans Affairs Desert-Pacific Mental Illness Research, Education, and Clinical Center (MIRECC); and for Peter Colvonen, PhD, by a Veteran Affairs Rehabilitation Science Research and Development Career Development Award (1lK2Rx002120-01). The authors report no conflicts of interest.

ABBREVIATIONS

AHI

apnea-hypopnea index

CPAP

continuous positive airway pressure

OSA

obstructive sleep apnea

PAP

positive airway pressure

PCL-5

PTSD Checklist for Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition

PHQ-9

9-item Patient Health Questionnaire

PTSD

posttraumatic stress disorder

RCT

randomized controlled trial

VA

Veterans Affairs