Drugs and obstructive sleep apnoea

We greatly appreciated the Letter to the Editor from Montastruc and colleagues 1 in response to our review dealing with the impact of concomitant medications on obstructive sleep apnoea (OSA) 2. As little is known about the putative role of drugs on OSA, we reviewed drugs in terms of their possible impact on OSA – i.e. drugs: (i) which may worsen OSA; (ii) which are unlikely to have an impact on OSA; (iii) for which data are scarce or contradictory; and (iv) with a potentially beneficial effect. The level of evidence was ranked according to three different grades: randomized controlled trials (RCTs) with a large sample size and sufficient power; RCTs with a small sample size, nonrandomized comparative studies and observational studies; retrospective studies or case reports. Our work has enabled us to propose expert clinical recommendations: drugs exacerbating OSA that should be avoided were identified (i.e. opiates, benzodiazepines, baclofen or drugs associated with weight gain), based on robust RCTs. The two pharmacovigilance studies investigating drugs associated with OSA from Linselle and colleagues 3, 4 offer very important additional data from original databases in real-life situations. These two studies use the French Pharmacovigilance Database (based on six spontaneously registered reports 3) and VigiBase®, the World Health Organization Global Individual Case Safety Report (ICSR) database (covering 3325 adverse drug reactions 4). The authors concluded that the main drugs associated with OSA were psychotropic agents (benzodiazepines, neuroleptic drugs), coxibs and opioids, reinforcing the results of randomized trials. Other pharmacovigilance signals have been described for drugs for which the underlying mechanism linking the drug to OSA was difficult to elucidate – concerning individual drugs (alendronic acid, digoxin and esomeprazole) and also classes of drugs (bisphosphonates, interferons, proton pump inhibitors and thiazolidinediones). The causal relationship with OSA remains to be demonstrated as these drugs might be markers of OSA-related symptoms or comorbidities rather than drugs favouring OSA or increasing its severity. For example, oesophageal reflux is clearly linked to huge variations in intrathoracic pressure during apnoeas and hypopnoeas, and it is plausible that proton pump inhibitors were used in this indication. Accordingly, a recent systematic review and meta-analysis concluded that proton pump inhibitors might improve the quality of sleep in OSA patients with concomitant gastroesophageal reflux disease, without any effect on apnoea–hypopnoea indices 5. Along the same lines, OSA is highly prevalent in patients with multiple sclerosis 6 and this might explain the pharmacovigilance signal observed for interferons. As an example of a contradictory result between the literature and the case–noncase study from Linselle and colleagues 4, the Cochrane Collaboration has proposed sodium oxybate as a potential therapeutic option for a subgroup of patients with OSA 7. We agree with the authors that quetiapine, clozapine, benzodiazepines, opium alkaloids and thiazolidinediones could worsen OSA. The weight gain induced by neuroleptic agents unequivocally worsens OSA 8. Finally, new pharmacovigilance signals have been provided by Linselle and colleagues for drugs previously unexplored in OSA – digoxin and bisphosphonates 4 – that need to be further investigated. It is important to note that our results were based mainly on patients, in whom there was well-documented sleep studies and a distinction between obstructive and central sleep apnoea. Information about polysomnographic testing is not systematically available in pharmacovigilance databases. Moreover, in most studies included in our review, patients had already been diagnosed with OSA before the introduction of the drug and the monitoring of apnoea–hypopnea index variations under treatment. This prospective approach differs from a case–non-case method, where reporting odds ratio (ROR) values have to be considered as a pharmacovigilance signal. It is now time to identify the specific mechanisms involved for each drug, or class of drug, and determine their true impact through well-designed studies. Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY 9, and are permanently archived in the Concise Guide to PHARMACOLOGY 2015/16 10. There are no competing interests to declare.