A novel medication approach to treating obstructive sleep apnoea

The paper by Thomson et al. (2025) in this issue of The Journal of Physiology is a novel and very interesting attempt to translate recent knowledge from the bench to the bedside. It reports the results of a placebo controlled double-blind cross-over trial of a novel drug, flupirtine, on 15 volunteers with known obstructive sleep apnoea (OSA). OSA is a common clinical entity affecting ∼20% of the adult population of most western countries. It is defined as the absence of breathing lasting for more than 10 s, occurring more than five times per hour during sleep. OSA syndrome includes a requirement for excessive sleepiness. OSA pathogenesis is considered to be a result of the complex interaction of respiratory control (controller gain), sleep state and airway collapsibility (Deacon & Catcheside, 2015; White, 2005).

Current main stream therapies for OSA include continuous positive airway pressure, mandibular advance devices, weight loss and upper airway surgery. In previous centuries some ‘tonics’ included small doses of strychnine, which raised muscle tone by blocking inhibitory spinal neurons. Over the years, many drugs have been tried as therapies for OSA (Hedner & Zou, 2018), including antidepressants such as mirtazapine to increase muscle tone, hypnotics such as benzodiazepines to reduce arousal threshold, progesterone to increase respiratory drive, and carbonic anhydrase inhibitors to reduce controller and plant gain.

Flupirtine is a KCNQ potassium channel opener (Costi et al., 2022) that was considered to potentially attenuate respiratory drive during sleep, thereby reduce the severity of OSA. The paper by Thomson et al. (2025) could be thought of as three experiments with flupirtine simultaneously. These comprise (i) the main study looking at overall effects on AHI and, in addition, two subsidiary ones trying to understand potential mechanisms of the expected (but absent) effect on apnoea-hypopnea index (AHI): (ii) a study on controller gain (Burgess, 2012) and (iii) a study aiming to assess airway collapsibility (Thomson et al., 2025).

Although there was no effect of the flupirtine on the severity of OSA in the group as a whole there did appear to be different effects in ‘good responders’ and ‘bad responders’. This dichotomy is well illustrated in their figure 4. In a subset of subjects, the AHI rose with drug (‘bad responders’ in red) and, in another subset, it fell (‘good responders’ in blue). There was no apparent correlation between those outcomes and the subject characteristics, nor during carotid body challenge testing to assess the degree of ventilatory responsiveness. One could interpret the results as confirming that the pathogenesis of OSA is indeed complex, but probably different in different patients. So, a drug that reduced controller gain might improve OSA in those patients with abnormally high drive, but not in those with normal or reduced drive. For them, the muscle relaxing properties of flupirtine made OSA worse.

Regarding the controller gain sub-experiment, their novel test method for carotid body stimulation (6% CO₂ in 14% O₂) was certainly a very powerful stimulus to the carotid body; the CO₂ stimulus was higher than we see in clinical practice. There may have been some stimulus to the central chemoreceptors as well, but that would not negate the results. There was no effect on controller gain by flupirtine. In retrospect, however, there was a good deal of scatter in the data as shown in their figure 3, as well as an effect on other subjects who all had high ventilatory drive may have yielded a positive result.

Regarding airway collapsibility, they have used an ‘in house’ approach for assessing collapsibility and drive, derived from manipulation of the flow signal in various degrees of hypoventilation. If we look at their figure 4, there were a few subjects where collapsibility appeared to be worse with flupirtine (in red) and where the arousal threshold appeared to be higher. That combination could explain why AHI was higher in REM sleep in some. Because there was no effect on controller gain, we can assume this was a result of the unrelated muscle relaxant and sedating properties of the drug (Costi et al., 2022) that caused those effects in some subjects. As they point out in their conclusions, this may worsen OSA in patients with co-existent drug-resistant epilepsy or other neuropsychiatric disorders (Thomson et al., 2025).

Thomson et al. (2025) went to extreme lengths to avoid potential confounders, which made the experiment and the subsequent analysis very complicated; however, the graphical illustrations have made the analyses understandable.

One could criticise Thomson et al. (2025) because they have not allowed for the multiple statistical comparisons. Hence, although the results of some the subgroup analyses are interesting and thought provoking, the results may have occurred by chance. Nevertheless, even negative studies should be reported, especially when they are well done, as this one was, because the results and discussion may stimulate others to revisit the issues in different ways. For example, other drugs that might influence controller gain (similar to acetazolamide), reduce arousal threshold (benzodiazepine derivatives) or increase muscle tone (analogues of mirtazapine, possible derivatives of strychnine), or make the upper airway less likely to collapse (weight loss drugs and artificial surfactant to reduce surface tension).