The distributions of SWA for three sleep cycles for control subjects vs untreated SAS patients and for treated vs untreated patients are presented in Figure 2, top and bottom, respectively. There was no interaction effect between group (control subjects and untreated patients) and NREM episode (1, 2, 3). However, an effect of NREM episode (F[2,30]), 13.1; Huynh-Feldt, p = 0.0001) was found, as can be seen in Figure 2, top.
A second ANOVA with two repeated measures (SAS patients before and after CPAP treatment and SWA in successive NREM episodes) showed an interaction between the two factors (F[2,12], 4.97; Huynh-Feldt, p = 0.027). To decompose this interaction effect, an analysis of simple effects was performed and showed a significant pretreatment to posttreatment difference for the first (p = 0.024) and second NREM episodes (p = 0.002); the difference for the third NREM episode was not significant.
As shown in Table 2, the mean sleep latency on the MSLT was significantly correlated with SWA in the first NREM cycle (r = 0.56; p = 0.045) before treatment. The microarousal index was significantly correlated (negatively) with the SWA in the first NREM episode and the total accumulation of SWA for the entire night. There was no significant correlation between the MSLT and either the percentage of REM sleep, the AHI, the Sa02 minimum, or the time spent with Sa02 < 90%.
One of the major difficulties in studying SWA in SAS patients results from the numerous artifacts associated with repetitive microarousals or awakenings closely related to respiratory impairments. To our knowledge, this is the first study of all-night SWA in patients with SAS, and there is no easy way and no validated or standard method to reject artifacts in this population.
We decided to exclude bursts of 8 activity occurring at the end of the apneic episodes in close association with microarousals, since it was previously reported that these bursts are part of an arousal response rather than physiologic SWA associated with the restorative functional sleep as SWA seen during SWS. One may question whether the criteria used for artifact rejection, including the rejection of “prearousal 8 waves” occurring at the end of apneic episodes, may have influenced the calculation of the total SWA across the night. To assess this possibility, we also calculated SWA across sleep cycles without rejecting these bursts of SWA. We obtained the same results. Correlations between EDS, as measured by the MSLT, and SWA in the first NREM-REM sleep cycle, also remained significant when the calculations were made without rejection of bursts of 8 activity occurring at the end of the apneic episodes.
Figure 2. Mean standardized time course of SWA across three NREM-REM cycles for 10 control subjects and 10 SAS patients before treatment (top) and for 10 SAS patients before and after treatment (bottom). There was a significant improvement in the level of SWA with CPAP treatment for the first and second NREM episodes. Note that for the before-treatment condition, three SAS patients had not completed their third cycle. However, it appears here for descriptive purposes.
Table 2—Pearson Product-Moment Correlations Between MSLT, SWA, and Sleep Disruption
|SAS Patients (Pretreatment)||SAS Patients (Posttreatment)|
|Correlations||r||p Value||r||p Value|
|MSLT and total SWA||0.30||NS||— 0.09||NS|
|MSLT and NREM 1 SWA||0.56||0.045||0.12||NS|
|MSLT and stage REM%||0.29||NS||0.27||NS|
|MSLT and AHI||0.09||NS||0.15||NS|
|MSLT and Sao2 minimum||— 0.39||NS||0.05||NS|
|MSLT and time with Sao2 < 90%||0.38||NS||— 0.06||NS|
|MSLT and microarousal index||— 0.04||NS||0.04||NS|
|Microarousal index and total SWA||— 0.78||0.004||— 0.48||NS|
|Microarousal index and NREM 1 SWA||— 0.64||0.024||— 0.54||NS|