Cardiovascular and inflammatory proteomic signatures distinguish mandibular advancement device from CPAP therapy and predict blood pressure response in obstructive sleep apnea

Background The the biological mechanisms underlying cardiovascular responses to obstructive sleep apnea (OSA) treatment remain unclear. We investigated whether inflammatory and cardiovascular proteomic response profiles differ between Mandibular advancement device (MAD) and continuous positive airway pressure (CPAP) OSA treatment, and whether circulating proteomic markers predict BP response. Methods CRESCENT randomized 220 patients with moderate-to-severe OSA and elevated cardiovascular risk to MAD or CPAP. Inflammatory and cardiovascular Olink proteomic profiles were measured at baseline and 6 months. In primary analysis, we compared 6-months protein levels between MAD and CPAP treatment. In secondary analysis, we estimated the effect of any OSA treatment and treatment adherence on 6-months protein levels, and baseline protein level associations with ambulatory BP outcomes at 6 months. Protein levels were rank-normalized jointly across the two time points. Analyses were adjusted for the baseline levels of the outcome as a covariates. False discovery rate (FDR) correction was applied. Results Follow-up at 6 months was completed by 89.1% of participants in the MAD group and 91.8% in the CPAP group. Overall, the median age is 62.0 years, 85.4% are male, median BMI is 27.7 kg/m2 and median baseline AHI is 39.8 events/hour. Median nightly device use was 5.5 hours for MAD and 5.0 hours for CPAP, with ≥4 h/night adherence achieved by 74.7% and 70.2%, respectively. In primary analysis, MAD therapy was associated with higher concentrations of six cardiovascular-related proteins compared with CPAP, including TFF3 (β = 0.176 standard deviations (SD) FDR = 0.016), uPA (β = 0.303, FDR = 0.016), EPHB4 (β = 0.176, FDR = 0.033), PI3 (β = 0.203, FDR = 0.035), IL-18BP (β = 0.179, FDR = 0.068), and TNF-R2 (β = 0.170, FDR = 0.093). Higher adherence was associated with lower concentrations of inflammatory proteins TNF (β = −0.001SD), IL-24 , and SIRT2, with consistent findings using a ≥4 h/night adherence threshold, regardless of treatment. Baseline cardiovascular proteomic profiles predicted BP responsiveness. Higher baseline OPN was associated with greater reductions in awake and 24-hour systolic BP, awake and 24-hour mean BP and 24-hour pulse pressure. Higher baseline U-PAR and uPA were associated with reductions in systolic BP and pulse pressure. Higher NT-proBNP was associated with reduction in asleep pulse pressure. All associations are independent of treatment assignment and adherence. Conclusions MAD therapy was associated with differential expression of selected cardiovascular-related proteins compared with CPAP, while baseline cardiovascular proteomic profiles predicted BP improvement. OSA treatment adherence is associated with inflammatory proteins.