Opioid-related sleep-disordered breathing: correlates and conundrums

The opioid crisis has brought to the fore the observation that opioid use is a distinct cause of sleep-disordered breathing (SDB). Prescription opioids have been associated with an increased prevalence of SDB ranging from 42%–85% [1] with a higher prevalence of central sleep apnea (CSA), 24%–35%, than in the general population. The respiratory patterns associated with opioids are distinctive, including ataxic, Biot’s or cluster breathing, and periodic breathing with short cycle time, unlike Cheyne Stokes Respiration [2]. This pattern is noted even at lower doses of opioids [3] (~20–50 mg morphine equivalent dose). Interestingly, this pattern is similar to the irregular breathing pattern observed after ablation of the pre-Bötzinger Complex in the rat model [4]. Furthermore, there is evidence that chronic prescription opioids are associated with severe non-apneic sustained hypoxemia, likely representing sustained hypoventilation and decreased gas exchange [5]. In this issue of SLEEP [6], Orr and colleagues aimed to delineate the underlying pathophysiology of SDB in relation to opioid use. The authors retrospectively performed signal analysis on polysomnographic studies (PSG) in 178 opioid users and 534 controls, matched on age, sex, body mass index, benzodiazepine use, heart failure, and atrial fibrillation. The goal was to determine whether one or more physiological endotypes, including loop gain, arousal threshold, compensation, respiratory rate, and the coefficient of variation of respiratory rate (RR), mediate the association between opioid use and PSG parameters. The PSG and physiologic parameters of interest were compared between opioid and control individuals using mediation analysis after adjusting for relevant covariates. The investigators used a previously validated PSG model-based tool of physiological traits underpinning SDB to estimate the effect of opioid use [7]. The authors found that opioid use mediated a higher loop gain, a lower respiratory rate, and a higher RR variability relationship, the latter during non-REM sleep. In contrast, there was no mediation effect of opioid use or the endotypes on hypoxemia. Additionally, central apneas were associated with higher loop gain, lower arousal threshold, higher compensation, and higher respiratory rate variability. Interestingly, there was no direct effect of opioids on the central apnea index. Prior studies in animal models have also indicated that opioids bind to the pre-Bötzinger Complex, the site of respiratory rhythmogenesis, resulting in an erratic ataxic breathing pattern [8]. While human studies have provided conflicting data on the acute [9–11] vs. chronic effects of opioid drugs on hypercapnic and hypoxic ventilatory responsiveness [12], all types of opioid drugs are associated with central apneas and unstable breathing [13], with a consistent feature of dose-dependent severity of opioid-related SDB [3]. Thus, the different components of opioid-related SDB must be examined in-depth when studying the underlying mechanisms of effect. The authors’ efforts of “peeling away the complex layers” of a mechanistic riddle are laudable. The study provides a systematic evaluation of the underlying pathophysiology of opioid-related SDB on available PSG studies using noninvasive tools. The analyses adjusted for multiple relevant covariates and relied on multiple mediation analyses to provide putative mechanisms of opioid effects on PSG parameters. However, many methodological considerations may influence the interpretation and generalizability of the stated findings, as noted below. First, as the authors acknowledge, the retrospective cross-sectional analysis disallows accurate opioid dose and duration estimation, night-to-night variability, and potential misclassification of patients into the opioid group versus the control group. This would result in a diminution of the strength of the mediation effect. Second, even if the prescribed opioid dose were available, serum dose levels rather than the dose estimated by oral morphine equivalent dose are deemed more accurate due to the variable bioavailability of the drugs [14, 15]. Furthermore, hypopnea classification as central or obstructive was not performed; hence, the analysis does not allow accurate estimation of central SDB in either group. The above constraints and the small effect size of loop gain, albeit statistically significant, may also potentially explain the absence of a direct effect of opioids on the CSA index in this study. Moreover, loop gain was not statistically significant in the moderate to severe SDB category, likely because the analysis was not powered to evaluate this association. Third, the noninvasive nature of the analysis enhances utility in a clinical setting. Conversely, certain gaps in physiologic parameters require detailed experimental measurements to be addressed definitively. For example, it is uncertain if hypoxic or hypercapnic chemoreflex sensitivity is adequately captured by PSG-based treatment of physiologic traits. Since loop gain ultimately depends on chemoreceptor modulations across a wide range of ventilation and chemical stimuli [2], loop gain measures derived from routine PSG require further validation using specific measures of chemoreflex sensitivity, both hypoxic and hypercapnic. Similarly, the lack of upper airway dilator muscle response is intriguing, given that opioids have been associated with decreased hyoglossus muscle tone in animals [16]. Potential confounders include body position, low opioid dose, inaccurate opioid dose estimation, and/or misclassification of patients. A definitive answer to the above question requires measurement of upper airway resistance using quantitative flow and supraglottic pressure measurement. Fourth, the finding of high RR variability with central apneas likely represents central arrhythmogenesis (simulating Biot’s breathing). Biot’s breathing was opioid dose-dependent in a prior study [3] and the correlation of RR variability with central apneas in this study, may represent an epiphenomenon of the opioid dose effect. Finally, a subpopulation of opioid SDB patients has hypercapnia [17]. In the absence of measurement of arterial blood gas or end-tidal carbon dioxide, it is not possible to determine whether hypoventilation may have contributed to the presence of hypoxia. The study by Orr et al. highlights the complexity of opioid-related SDB, necessitating the development and testing of individualized approaches guided by pathophysiology. Future studies could address whether elevated loop gain is due to increased controller gain (i.e. chemoreflex sensitivity) or increased plant gain due to centrally mediated hypoventilation. The co-occurrence of central and obstructive events suggests that combined therapy that includes continuous positive airway pressure [18] and another intervention may be warranted. Interventions that may decrease plant gain include positive airway pressure therapy or acetazolamide [2]. Increased chemoreflex sensitivity seems unlikely, but it was found in one study in stable methadone maintenance treatment patients. Supplemental oxygen has been shown to ameliorate central apnea in older adults and to decrease the propensity to CSA via decreasing loop gain [19]. Innovative therapies targeting rhythmogenesis should also be explored to alleviate this condition [20]. The authors are congratulated on addressing an important public health issue and for confirming, in a real-world setting, the myriad pathways affecting SDB in this population. They have clearly demonstrated the need to develop and test targeted interventions underpinned by a robust understanding of the pathophysiologic mechanisms. Conversely, this also highlights that decreasing the impact of opioid use remains the most effective way to mitigate the scourge of opioid-related SDB and the tragic individual and societal impact. This work was supported by VA CSRD Merit Review Award Grant number: 1I01CX001938-02A0. Financial disclosure: The authors have nothing to disclose. Nonfinancial disclosure: The authors have no potential conflicts of interest.