The present meta-analysis showed that CPAP therapy improves serum levels of the inflammatory markers CRP, TNF-α, and IL-6. Using a p-value of <0.05 to mark a significant change, the levels of CRP and TNF-α were significantly decreased, whereas the levels of IL-6 showed no significant change. IL-6 levels did, however, show a general trend of decreasing values with CPAP usage.
With respect to CRP, the following studies agreed that CPAP usage significantly decreases serum levels of CRP: Iesato et al. 2007 , Ishida et al. 2009 , Patruno et al. 2007 , Schiza et al. 2010 , Steiropoulos et al. 2007 , Yokoe et al. 2003 , and Zhao et al. 2011 . With respect to TNF-α, the following studies agreed that CPAP usage significantly decreases serum levels of TNF-α: Minoguchi et al. 2004 , Ryan et al. 2005 , Ryan et al. 2006 , Steiropoulos et al. 2009 , and Tamaki et al. 2006 . With respect to IL-6, the following studies agreed that CPAP usage significantly decreases serum levels of IL-6: Burioka et al. 2008 , Ye et al. 2010 , and Yokoe et al. 2003 .
We also examined why some studies did not agree with our overall finding that CPAP significantly improves levels of inflammatory markers. With regards to CRP, Carniero et al. 2009  showed there was no significant change in the levels of CRP after CPAP, and this could possibly be due to the small sample population (7 subjects) of the study. Chung et al. 2011 , Dorkova et al. 2008 , Harsch et al. 2004 , and Ryan and colleagues  showed no significant change in serum CRP levels, possibly because of the small sample populations (25, 16, and 20, respectively) and possibly because there was no significant weight reduction among the patients in the study. There is some debate on whether CRP levels are dependent on obesity or the severity of OSA [31, 81]. Kohler and colleagues  performed a randomized controlled trial and similarly concluded that 4 weeks of CPAP had no significant reduction in CRP levels, possibly due to the fact that many of the subjects also had a number of other comorbidities in addition to OSA.
Five of the nine studies measuring TNF-α levels showed that CPAP usage significantly decreases serum levels of TNF-α [25, 60, 75–77]. Studies that showed no significant change in TNF-α were examined to determine why those studies did not agree with our overall findings. Carniero et al. 2009  had a very small sample population (7) and showed that CPAP usage over 3 months decreases serum TNF-α, though not significantly. Guasti et al. 2009  also had a small sample population (only 16) and many of the patients had other comorbidities, such as elevated BMI. In addition to a small sample population (16) and patients having high BMIs, Vgontzas and colleagues  noted that there were discrepancies in the compliance of the patients using CPAP. Only 10 of the patients used CPAP for more than 4 hours per night.
With respect to IL-6, only three of nine studies showed that CPAP usage decreases serum levels of IL-6 [18, 79, 80]. The rest of studies examined showed no significant change in serum IL-6 after CPAP treatment. Again, these studies all had populations under 50 people, and the patients also exhibited comorbidities, like obesity. A few of these studies [18, 76, 82] only measured the effect of CPAP on systemic inflammation over 4–6 weeks, which is relatively short compared to many of the other studies examined.
There are a few limitations of this meta-analysis. It is very clear that the available literature is largely low-level evidence. Most of the studies included in the meta-analysis have examined the confounding factors (age, AHI, BMI), which we did not adjust, since we did not perform a meta-regression analysis. Moreover we did not perform the subgroups analysis to examine effect of severity of OSA on inflammatory markers before or after treatment. There are number of studies available in which levels of these markers were measured in patients with OSA and controls. Those studies cannot be included because of significant methodological differences (no CPAP). We performed a metaregression analysis on this larger pool of studies (submitted to JCSM being reviewed). Also, we did not account for CPAP compliance rates. If we had included the data from non-compliant CPAP groups, the serum levels of the selected inflammatory markers may have been affected. Another potential limitation is that we excluded all papers written in languages other than English, which could raise the possibility of publication bias. We have excluded some studies with exponentially high values when compared to the other studies in this meta-analysis. Including those studies could affect the net result of the meta-analysis. We were not able to retrieve the numerical data from some studies that reported data only in graphical form. That data inclusion could have affected the results of our meta-analysis. It is known that studies with positive results tend to get published while studies with negative results are less likely to be published, and we only included data from published studies in our meta-analysis. This could have led to publication bias as well.
Despite all these variations, it was reassuring that in majority of the studies (regardless of the composition of the study) those with CPAP treatment have lower levels of systemic inflammatory markers. This suggests that selection and sampling biases were unlikely to be responsible for the observed associations. The improvement in inflammatory markers suggests that OSA treatment modulates the cardiovascular risk profile through multiple mechanisms, including inflammation, which may play an important role for the development of atherosclerosis. Further studies are required to explore this dimension of the cardiovascular risk profile, such as the impact of OSA treatment on atherosclerosis and vasculopathy.
In conclusion, CPAP usage for patients with OSA significantly decreases serum inflammatory markers CRP and TNF-α. Also, CPAP usage seems to decrease serum levels of IL-6.