Through visual examination of EEG recorded during LOC, we noticed several events similar to slow waves seen during natural sleep. These events occur almost in all LOC periods, usually once a second, which is equivalent to the occurrence of slow waves in non REM sleep. Wave only occurred during LOC (Ramsay score 5-6, 3.87 ± 1.39 mcg/mL propofol, see supplement). Each event can be seen on a large number of channels. The frequency and spatial distribution of events are similar (average 107.9 channels/wave, from 87.4 channels/wave to 124.8 channels/wave). Then, we use an automatic detection process (see Methods and Reidner et al.) to separate a subset of these events. For our 8 subjects, we detected 295 waves. This similarity was confirmed by comparing the events recorded during propofol anesthesia with the population whose slow wave amplitudes were matched during spontaneous NREM sleep recorded by 6 other subjects. We found that propofol events have a similar scalp voltage topography to spontaneous sleep slow waves. The comparison of terrain with negative and positive slopes shows that, as in spontaneous sleep, the maximum slope appears in a cluster of frontal central electrodes (E). Except that one channel showed more negative slopes in propofol LOC, there was no statistically significant difference in any slow wave parameters.

In a recent paper, we used hd-EEG and source modeling to analyze the cortical current under slow wave sleep in individuals. Here, we use the same technique to examine a single propofol wave. Like sleep slow waves, each propofol slow wave is a unique and spreading cortical wave. Figure 3 shows two representative propofol waves. The wave on the left originates from the left insular cortex, and then spreads forward to the bilateral frontal cortex. At the end of the wave, the island current weakened. Another wave spreads forward from the origin of the posterior cingulate, and the origin of the posterior cingulate remains strongly activated during the wave duration.

For each propofol slow wave, we measured three parameters: origin, propagation, and intervention. The origin is a group of cortical voxels that may contain slow wave initiation (see method). As previously reported on spontaneous sleep, propofol slow waves often originate from hot spots in the insular and cingulate cortex. We use the time of the current peak in each voxel to generate the gradient. Streamlines along the gradient of the origin voxels describe the propagation of waves through the cortex. We calculated the frequency of slow wave propagation of propofol through each voxel. We found high-density transmission in several midline structures including anterior cingulate, cingulate and posterior cingulate. These same structures constitute the backward propagation of the spontaneous slow wave and anesthetic slow wave of the middle expressway, which were previously reported in the spontaneous sleep slow wave, preferentially along the expressway. We define participation as the magnitude of the current generated in each voxel during the wave (see method). Propofol slow wave is associated with large currents in many of the same areas of spontaneous sleep slow wave, including anterior cingulate, posterior cingulate, and anterior cuneiform lobe.

Propofol slow wave is different from sleep slow wave

Although propofol slow waves are very similar to those observed during natural sleep, there are two distinct differences. First, spontaneous sleep slow waves have been shown to group spindle (12-15 Hz) activity, and spindle power increases when the slow wave is tilted in the forward direction. Here, we measured the root mean square error of window spindle activity centered on the negative peak of each sleep and propofol slow wave. We found that there was almost no grouping between the slow wave of propofol and the root mean square difference of spindle activity. During spontaneous sleep, the root mean square of spindle at the slow wave forward slope increased compared with that before and after the forward slope (P0.05, Student t-test). This is not the case with propofol slow waves.

Secondly, evidence from EEG and source level analysis shows that propofol slow waves are spatially ambiguous compared with spontaneous sleep slow waves. For EEG parameters, compared with spontaneous sleep, the difference between electrodes when taking propofol decreased in both positive and negative slopes (unpaired t-test, P0.05), and the average terrain also tended to this direction (unpaired t-test, P0.08). At the source level, the propofol slow wave contains significantly more cortical voxels than the amplitude matched spontaneous sleep slow wave population (unpaired t-test, P0.05, see Method,). In addition, the origin hotspots of propofol slow waves are more scattered, and some parts of the occipital cortex that rarely originate in spontaneous sleep have multiple origins in LOC. Areas such as the posterior temporal lobe and occipital cortex were mostly ignored in spontaneous sleep slow waves, but showed moderate activation in propofol slow waves. Finally, during spontaneous sleep, there was a statistical asymmetry between the left and right frontal cortex (Student t test, P0.05), but not in propofol slow wave.