Introduction

Both opioid intoxication and withdrawal disrupt sleep architecture at a physiological level [1, 2]. At the same time, sleep disturbances themselves predispose individuals to opioid use disorder (OUD) [3–5]. A growing body of literature indicates that neurobiological, neuropsychiatric, and social-ecologic factors might mediate the bi-directional relationship between sleep disturbances and OUD [4, 6]. These are important relationships to understand because sleep disturbances (e.g., insomnia) are common among individuals receiving medication for OUD (MOUD), like buprenorphine [7], and are associated with negative OUD treatment outcomes, including substance use recurrence. Thus, strategies to improve sleep may be a novel approach for improving treatment outcomes for people with OUD [6, 8].

Previous evidence indicates that poor sleep contributes to the development and maintenance of substance use disorders by heightening pain sensitivity, promoting negative affect, and adversely impacting stress reactivity as well as self-care-related executive functions [1, 4, 9–11]. These relationships may arise via a depletion of “cognitive resources” [12] or disruption of neurobiological systems important for arousal, such as imbalance in the orexin/hypocretin system [3, 13, 14]. In humans, these disruptions can manifest as differences in incentive-related decision-making that can be measured using validated metrics [9, 13, 15–17]. Moreover, sleep quality can impact how individuals choose to respond (e.g., to avoid) exertion of effort [16], which could be a mechanism for poorer performance on a variety of tasks. Sleep quality may thus influence a set of neurofunctional domains that are believed to predict substance use treatment outcomes [21]. However, previous studies have typically considered only a narrow set of neurofunctional indices within the context of a single design or relied on samples whose subjects were actively engaged in substance use. Thus, the relationship between sleep and broad neurofunctional outcomes in addiction treatment samples is not well characterized.

Here, we attempt to bolster the existing literature by examining the prevalence of insomnia and its neurofunctional correlates in the domains of reward, cognition, interoception, negative emotionality, and metacognition among a sample of individuals receiving buprenorphine for OUD. Sleep disturbances such as insomnia may alter the underlying neurofunctional mechanisms of OUD at various stages of the OUD treatment and recovery. Obtaining insights on how sleep disturbances relate to other symptomatology would be useful in the creation of individualized treatment plans for patients with OUD and comorbid insomnia by identifying potential targets for intervention [9] such as adjunctive pharmacologic (e.g., orexin receptor antagonists, antidepressants, etc.) and non-pharmacologic (e.g., CBT-I, working memory training, cue exposure therapy, etc.) treatments to medications for OUD. Given the multi-factorial (e.g., genetic, environmental, lifestyle) etiology of both sleep disturbances and substance use behaviors [18], the capacity for personalization in treatment for this population represents a promising avenue to improve OUD outcomes [1, 19, 20].

Materials and methods

Recruitment and testing procedures of this protocol were approved by the Virginia Commonwealth University Institutional Review Board.

Results

Sample description

Overall, respondents (N = 129) identified predominately as cisgender women (62%), white (55.8%), unemployed (44.2%), and on a sublingual buprenorphine dose of ≥16 mg/day (47.3%). Roughly 60% of participants reported clinically significant insomnia symptoms. Due to the group definitions adopted here, the mean (standard deviation [SD], range) ISI score for the “minimal evidence of insomnia symptoms” group (INS-) was 5.26 (3.3, 0–10) and for the “evidence of insomnia symptoms” group (INS+) was 18.72 (4.6, 11–28). On average (SD, range), participants in this sample had started buprenorphine therapy 2.52 (2.30, 0.12–13.20) years before the time of data collection. Demographic characteristics did not significantly differ between the two insomnia symptom status groups (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Sociodemographic and key clinical variables by insomnia symptom status group.

https://doi.org/10.1371/journal.pone.0304461.t001

Neurofunctional correlates of insomnia symptoms

For neurofunctional measures (Table 2), within the negative emotionality domain, INS+ participants were more likely to endorse symptoms of poorer overall mental health, depression, anxiety, and post-traumatic stress than INS- participants (ps<0.001). Moreover, the INS+ participants were more intolerant to stress, exhibited lower acceptance of stress and trust in stress-coping skills, as well as reported being more absorbed by distressing emotions (ps<0.01). In the Metacognition and Interoception domains, evidence of clinically significant insomnia was associated with having negative beliefs about uncontrollable and dangerous thoughts (p<0.05) as well as heightened awareness of bodily sensations (p<0.01). No statistically significant differences were observed within the reward and cognition domains between INS groups (ps>0.05). Results were insensitive to imputing missing values (S2 Table in S1 Appendix), controlling for possible demographic confounders (S3 Table in S1 Appendix), adopting a higher ISI cutoff score (S4 Table in S1 Appendix), or modeling ISI scores as a continuous variable in covariate-adjusted regressions (S5 Table in S1 Appendix).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Unadjusted differences in key neurofunctional measures as a function of insomnia status group.

https://doi.org/10.1371/journal.pone.0304461.t002

Discussion

Our findings indicate that among a clinical sample of individuals with OUD receiving buprenorphine, clinically meaningful insomnia symptoms are a prevalent issue and one identified by patients as a barrier to OUD treatment progress. Moreover, patients who experienced a higher burden of insomnia symptoms also reported significantly higher levels of depression, anxiety, post-traumatic stress, stress intolerance, unhelpful metacognition, and interoceptive awareness. Participants experiencing a higher burden of insomnia symptoms in this sample also showed potentially meaningful increases in aspects of impulsivity, such as “negative urgency” (i.e., tendency to act rashly when distressed), though between group differences in these measures did not achieve statistical significance. Our results highlight how insomnia symptoms can persist in patients with OUD receiving buprenorphine for years, especially when not addressed directly.

Because the neurofunctional correlates of insomnia symptom severity identified here are also associated with addiction treatment outcomes [9, 10, 37], we speculate that they represent a mechanism through which poor sleep contributes to an increased risk of return to substance use in OUD patients. Thus, although correlative, these data indicate that combining MOUD treatment with strategies that address insomnia symptoms might help to improve treatment outcomes. For example, pharmacological (e.g., orexin receptor antagonists) or behavioral (e.g., cognitive behavioral therapy for insomnia) interventions could be leveraged in the design of personalized treatments for patients with OUD alongside medications for OUD (i.e., buprenorphine). Given the diverse ways in which sleep disturbances can arise and impact substance use behaviors, developing personalized treatment strategies in this OUD populations could be a powerful approach to improving outcomes by considering the unique barriers to recovery that individual patients may face. Studies testing this possibility would also shed light on the causal relationship (and directionality) between sleep dysregulation and OUD treatment outcomes.

Our results are largely consistent with previous literature linking poor sleep quality to negative affect, stress reactivity, and pain sensitivity, but expand knowledge by drawing upon a diverse outpatient population currently receiving MOUD and utilizing a multidimensional assessment of neurofunction. Our results also synergize with previous literature describing elevated autonomic responses to relived distressing experiences among people with insomnia, suggesting that increased experiences of negative emotionality and struggles with managing the resulting stress are potentially important consequences of insomnia in OUD treatment populations [38]. Deficits in neurofunction identified using emotional data are a potentially distinct consequence of insomnia [39] from deficits in other areas of neurofunction such as procedural learning, attentiveness, and executive functioning [40]. The insomnia-emotionality link may be especially salient in substance use populations, which tend to have high rates of previous trauma and comorbid psychiatric conditions [41, 42]. Additionally, the neurofunctional differences by insomnia symptom status identified in this study largely align with the domains represented in the predominant neurofunctional phenotype our group recently [43] identified among a large sample of individuals with addiction using exploratory factor analysis of NIDA PhAB data (i.e., sleep, negative emotionality, metacognition, reward). Although not studied here, it is interesting to note that the orexin/hypocretin system is a common regulator of each of these functional outcomes [44, 45] and may therefore serve as a common biological mediator of the effects observed here. Indeed, orexin neurons are upregulated in response to morphine and heroin [46] which is expected to have important implications for sleep, stress and cognitive outcomes [47]. Overall, these findings highlight how sleep disturbances may be part of a constellation of symptoms with substantial neurofunctional overlap in patients with OUD, possibly via neurobiological interconnections such as the orexin system—future work is required to test this hypothesis [13, 14].

These results should be considered in light of several limitations. The use of a cross-sectional survey meant that we were unable to obtain specific evidence that the sleep disturbances in the INS+ group played a causal role in the emergence of other symptomatology; future studies should seek to test if changes in insomnia symptoms (including improvements associated with pharmacological or psychological interventions) are related to changes in neurofunctional phenotype and OUD treatment outcomes over time. Investigations of the patterns of comorbidity of mental illnesses [48] and their shared genetic [49] and neurodevelopmental [50, 51] architecture point to the existence of a common “p-factor” that nonspecifically indexes overall mental health burden in an individual, where sleep disturbance is thought to be a component of the p-factor [52]. By extension, poor sleep in the more symptomatic MOUD patients could simply be a manifestation of this; in this case, sleep interventions may be expected to have less impact compared to a scenario where sleep directly modulates mental health/addiction outcomes. In addition, due to exclusion of patients with severe comorbid mental illness, our findings may not generalize to other treatment contexts, populations, or sleep disorders. However, it is likely that inclusion of MOUD patients with more psychiatric comorbidity (i.e. higher p-factor) could have yielded even stronger results. The null results in the “cognition” domain of the PhAB-Brief documented here may be attributable to our use of a temporal discounting task instead of an effort-based discounting task [16]; future work should explore differences in discounting task designs and seek to understand how insomnia may differentially impact each of these impulsivity measures. Participant responses on some items may have been impacted by recall and social desirability biases [53] and the relatively small sample size limits the generalizability of our findings. Finally, these analyses did not correct for family-wise errors which raises the possibility of having reported some false-positive results. Results reported here should thus be considered as “hypothesis-generating” rather than confirmatory. Despite these limitations, our findings may help explain why patients with sleep disturbances experience worse OUD treatment outcomes than their counterparts. Our data also highlight the need for future work to identify biological systems that might mediate sleep dysregulation during OUD treatment, which might serve as targets for novel treatments to improve treatment outcomes.

Conclusion

We report that the subset of MOUD patients with sleep problems (i.e., insomnia) in the clinically-significant range show worse negative affect symptomatology, poorer frustration tolerance, and other features not conducive to OUD recovery. This suggests that the link between impaired sleep and psychiatric illness severity found in other clinical contexts and populations is also evident in the MOUD population. Improved understanding of the linkages between sleep disturbances, neurofunction, and substance use behaviors may be useful in the development of novel, individualized treatment strategies for individuals with OUD and comorbid insomnia. Future studies could expand upon these findings by measuring changes in sleep, neurofunction, and substance use behaviors over time (such as after interventions that target sleep), as well as by deploying neuroimaging and other modalities to undercover mechanisms of shared symptomatology between sleep and collateral problems in SUD populations. Finally, our data suggests that insomnia could be a symptom target in future clinical trials in patients with opioid use disorder in treatment with buprenorphine.

Supporting information

Acknowledgments

We would like to thank the participants of this study for sharing their time and insights with our team.