Sleep Matters

Sleep Matters

Revisiting oxygen for hypertension in OSA

Author: Omar Mesarwi, MD
Date: March 30, 2019

Ah, summer 2014. Seems like just yesterday that Sam Smith was hitting the high notes on “Stay With Me,” and John Legend was letting us know how much of him loved us (hint: all). But a hugely influential paper in sleep medicine also came out in the New England Journal of Medicine in June 2014, “CPAP versus Oxygen in Obstructive Sleep Apnea.” In this paper, Gottlieb and colleagues noted that since so many of the adverse cardiovascular effects of OSA seem to be correlated with the severity of nocturnal hypoxemia, it made sense to see if treating just the hypoxia component of OSA without the other effects that CPAP has (improving intermittent hypercapnia, thoracic pressure swings, and fragmented sleep, among others) would result in an improvement in blood pressure as a primary outcome. The authors also assessed various inflammatory markers and metabolic parameters as secondary outcomes.

The long and short of it was that while CPAP significantly reduced nocturnal and 24-hour blood pressure (similar to previous studies), nocturnal oxygen alone had no effect. This study seemingly put a nail in the coffin of efforts to use oxygen as “salvage” therapy for patients with OSA who couldn’t tolerate CPAP, and it suggested that the effects of CPAP on adverse health outcomes associated with OSA were diverse and not limited to improving oxygenation. As usual, there were limitations with the study, though: Patients with the worst OSA, and with the most significant hypoxemia, were excluded from the study. Moreover, oxygen flow was fixed at 2L/min irrespective of the depth of desaturations during respiratory events.

One interesting change in the OSA research landscape in the last few years has been that CPAP withdrawal has gained favor as an OSA model in clinical studies. In considering the effect of CPAP on whatever outcome you’re interested in, you might randomize people to CPAP or nothing (or sham), but a consistent issue with this approach has been that CPAP adherence is traditionally low. Subgroup analyses of those who are adherent to CPAP are also problematic, since you don’t know offhand whether there are particular traits about CPAP-adherent subjects which make them “different” from non-adherent subjects in some way. (Are they more likely to take their other medications, for example?) Using a CPAP withdrawal model, you eliminate most of this issue. You take patients who are already diagnosed with OSA, and already adherent to therapy. Then you can take them off CPAP and see what happens, effectively turning each subject into his or her own control. Jonathan Jun has found that CPAP dynamically increases lipid levels at night using a CPAP withdrawal model, and others have applied this in various ways as well.

And that’s exactly what the authors of the current study did too. Subjects with moderate to severe OSA and good CPAP efficacy and adherence (residual AHI <10 events/hr, use of 4 hours/night or longer for at least 1 year) had CPAP withdrawn for 2 weeks to gauge baseline blood pressure. Then they were randomized to getting either nocturnal oxygen or air at 5L/min, and blood pressure and nocturnal oximetry were recorded. Finally, after a two-week “washout” period on CPAP, the subjects were crossed-over to the other arm for another two week period. Blood pressure change was the primary outcome. Baseline characteristics were pretty similar between groups. Subjects who got air as an intervention had a pretty significant increase in blood pressure: 6.6/4.6 (sys/dia) mmHg. As expected, oxyhemoglobin desaturation index in the oxygen arm was significantly lower (6.4 events/hr versus 32.5 events/hr). Most importantly, supplemental oxygen essentially abolished the rise in morning BP during CPAP withdrawal.

This is a well-done study, which uses a great emerging model to give us new insight about the role of intermittent hypoxia in OSA-induced hypertension. Finally we have a more direct human correlate for the now decades-old (!) data from Eugene Fletcher’s group in rodents, where they showed that intermittent hypoxia causes hypertension. One limitation is that 24-hour blood pressure wasn’t assessed – it would be nice to see what happens to diurnal variations in blood pressure. Still, this is a nice study which gives us some new mechanistic insight about the role of hypoxia in human OSA. It’s worth mentioning, too, that although the absolute numbers in this study are still very much in the single digits, a 6 mmHg reduction in systolic BP is associated with a 15-20% reduction in coronary heart disease risk. That’s a pretty big deal on a population level.