Paradoxical decrease in isoprostane and increase in superoxide dismutase following CPAP withdrawal in OSA

To the Editor:

Stradling et al. [1] have recently published an interesting, unprecedented study assessing the effects of a continuous positive airway pressure (CPAP) withdrawal on oxidative stress in patients already treated for moderate-to-severe obstructive sleep apnoea (OSA). Based on a combination of two different studies, the authors showed that stopping CPAP for 2 weeks (“sham-CPAP” group) significantly decreased the urinary F2-isoprostane concentration, while it increased the activity of the blood superoxide dismutase (SOD), compared with uninterrupted treatment (“CPAP” group). The fall in urinary F2-isoprostane concentration observed in patients who returned to OSA was contrary to the hypothesis tested by the authors, i.e. an increase in oxidative stress when the disease returns. Interestingly, these paradoxical results are corroborated by the strong negative correlation they found between the re-increased oxygen desaturation index values after 2 weeks of CPAP withdrawal and the fall in urinary F2-isoprostane concentration. The authors advanced the hypothesis that intermittent hypoxia for 2 weeks of CPAP withdrawal may have caused a preconditioning protection against oxidative stress.

The fact that a CPAP withdrawal decreased this urinary biomarker of lipid peroxidation is quite unexpected and somewhat questionable. Indeed, on the whole, isoprostane production has been shown to be positively associated with sleep apnoea severity and degree of nocturnal hypoxia, even if some studies did not show such links. Whatever the biological fluid, most studies have shown increased concentrations of isoprostanes in OSA patients compared with control subjects. Above all, CPAP treatment for at least 2 months is effective in decreasing isoprostane concentration in urine, serum, plasma or exhaled breath condensate [2–4]. Consequently, because of their counterintuitive results, Stradling et al. [1] should provide some additional information to reinforce the advanced hypothesis. The first point is about the expression of the urinary F2-isoprostane results. Since renal dysfunction may impact the urinary concentration of biomarkers, urinary F2-isoprostane results should be expressed as pg per mg of urinary creatinine, as usually done [2], and the statistical tests should be reassessed. Furthermore, since intra- and inter-individual variations are known to be consequential for mean concentrations of urinary F2-isoprostane [5] and given the wide standard deviations reported by Stradling et al. [1] it would be more appropriate to show the new results graphically, ideally using a dot-and-line graph detailing each patient. Similarly, superoxide quenching by SOD results from its enzymatic activity, thus SOD measurement in Units·L⁻¹ would have been preferable, as usually expressed, and since mass concentration does not augur enzymatic activity [6]. The second point is about the method used for urinary F2-isoprostane quantification, i.e. the ELISA kit. Indeed, even if used in many clinical studies, the ELISA method is not the most reliable approach since different kits may give discrepant results [7]. Moreover, the results cannot be compared with those obtained by mass spectrometry, which remains the reference method. The last point is about the pathophysiological mechanism. It has been well-demonstrated in many cell models, animal studies and human diseases that SOD is activated under hyperoxic conditions rather than hypoxia, to counteract superoxide production [8, 9]. Nevertheless, the hypothesis that intermittent hypoxic preconditioning may protect from subsequent ischaemia by increasing antioxidant defence cannot be ruled out. To our knowledge, even if such preconditioning has been demonstrated for SOD activity in cell models, or in renal tissues from rats as cited by the authors [10], it has never yet been shown at a circulating level in human studies. It should be noted that in this reference cited as support, lipid peroxidation was not decreased after such preconditioning, as precisely assessed by renal isoprostane and malonedialdehyde concentrations [10]. Moreover, studies evaluating circulating SOD before and after CPAP in OSA patients are too rare and conflicting, perhaps due to the differences between the subtype of assayed SODs (cytosolic or extracellular Cu,Zn-SOD, and mitochondrial Mn-SOD), or because of pre-analytical and analytical biases regarding SOD measurements. Thus, as with the fall in urinary F2-isoprostane concentration, an increase in SOD activity after 2 weeks of intermittent hypoxia is also debatable. Isoprostanes are oxidised compounds formed in vivo via a non-enzymatic mechanism involving the peroxidation of arachidonic acid by overproduction of reactive oxygen species, whereas SOD acts as a major antioxidant by catalysing superoxide transformation into oxygen plus hydrogen peroxide through a dismutation reaction. Thus, given the fall in urinary F2-isoprostane concentrations (∼30%) and the rise in SOD ones (∼30%) one may expect concentrations of both biomarkers to be inversely correlated, which should be analysed by the authors.

To conclude, the deleterious role of oxidative stress in the cardiovascular consequences of OSA has been well-demonstrated. Most studies seem to indicate that CPAP enables lowering of this phenomenon. Consequently, the paradoxical decrease of lipid peroxidation and the rise in antioxidant enzymatic defence observed after a 2-week CPAP withdrawal period are quite unexpected results. This deserves to be reinforced before being reinvestigated in further studies using such interesting approach of treatment withdrawal.