Introduction

Precisely estimating the effects of general anesthetics is advantageous to decreasing the incidence of intra-operative awareness1, maintaining the hemodynamics more stable and reducing the requirements for intra-operative general anesthetics, fluids and vasopressors2. It is necessary to continually estimate the effects of general anesthetics during anesthesia to avoid excessively deep or light anesthesia. Because auditory evoked potential is easy to be recorded and sensitive to general anesthetics3,4, it has been widely used as a measure of the effects of general anesthetics5,6. Generally, AEP is characterized by its latency (the time from stimulus beginning to the AEP wave peak) and amplitude (the voltage of the wave peak). AEP latency monotonically increases in a concentration-dependent manner during anesthesia7,8,9,10,11,12,13, while AEP amplitude can decrease7,8,9,10,11,12,13, increase14 or change in multiple manners during anesthesia15. These results suggest that AEP latency is superior to amplitude for reflecting the effects of general anesthetics. It is well known that general anesthetics have effects on both nerve fibers16,17,18 and synapses16,19,20. However, it is unknown whether or not the AEP latency change during anesthesia can reflect the effects of general anesthetics on nerve fibers and synapses.

Although some auditory neurons exhibited paradoxical latency shifts (i.e., the spike latency increases with acoustic intensity increase) in bats21,22 or a much smaller/independent latency change with acoustic intensity in bats and gerbils23,24,25, the first spike latency decreases as the acoustic intensity increases (i.e., the latency-intensity curve) in the inferior colliculus (IC) of mice and cats26,Fig. 2f, the cyan curve, R2 = 0.147). The fitting resulted in the residual of each data. The residual, i.e., the difference between the actual value and the regression predicted value, can reflect the degree of variation of each data point. The absolute values of residuals obtained by fitting normalized A80-time curves were larger than those obtained by fitting normalized L80-time curves (Fig. 2g, 2 Independent Samples Tests, Z = −12.519, P = 5.890 × 10−36). Therefore, the AEP L80 from all mice changed with more regularity and less variability than the AEP A80 during pentobarbital anesthesia.

Changes in AEP latency and amplitude to various acoustic intensities during anesthesia

Although AEP L80 exhibited more regular and less variable changes than A80, whether AEPs to acoustic stimuli at other intensities undergo similar changes during anesthesia is still unclear. Therefore, we applied multiple acoustic intensities other than 80 dB to the subjects. The latency to each acoustic intensity (No. 18 mouse (M20110417)) showed an initial increase followed by a gradual decrease during anesthesia (Fig. 3a). The maximal latency to each acoustic intensity always occurred in the same recording session (Fig. 3a, the vertical green line). The latency change ranges (the maximal latency—the minimal latency) to different acoustic intensities were, respectively, 13.475, 12.344, 8.935, 6.23, 6.147 and 5.41 ms from 40 dB to 90 dB at 10 dB intervals and were negatively correlated with the acoustic intensity. These results indicated that AEP latencies to low intensity acoustic stimuli were more suitable for reflecting the effects of general anesthetics.

Figure 3
figure 3

Changes in AEP latency and amplitude to different acoustic intensities.

(a,b) Changes in latency and amplitude to 40–90 dB acoustic stimuli with anesthesia time, hereafter, latency- and amplitude-time curves. (c and d) R2 values and absolute values of residuals obtained by fitting normalized latency- and amplitude-time curves (computing methods were similar to those in Fig. 2) (*P < 0.05, **P < 0.01, 2 Independent Samples Tests).

Full size image

Although the amplitude (No. 18 mouse (M20110417)) also increased and then decreased during anesthesia to a given intensity acoustic stimulus, less regularity was found (Fig. 3b). Generally, the amplitude change to higher acoustic intensity was at a higher level (Fig. 3b. the curves for 40–80 dB). However, the amplitude change to a 90 dB acoustic stimulus (Fig. 3b, the black line with squares) decreased paradoxically when compared with those to 70 and 80 dB (Fig. 3b, the blue line with triangles and red line with circles). The maximal amplitudes to different acoustic intensities occurred in different recording sessions (Fig. 3b). Furthermore, the amplitude change ranges (the maximal amplitude—the minimal amplitude) to different acoustic intensities were, respectively, 0.132, 0.202, 0.137, 0.148, 0.16 and 0.178 mV from 40 dB to 90 dB at 10 dB intervals and showed little regular change with acoustic intensity increase.

When we applied the same methods as those in Fig. 2e to latency- and amplitude-time curves in Fig. 3a,b, at any acoustic intensity, the R2 value of fitting the normalized latency-time curve was larger than that of fitting the normalized amplitude-time curve (Fig. 3c). The fitting R2 values at low intensity acoustic stimuli (40, 50, 60 and 70 dB in latency; 40 dB in amplitude) were larger than those at high-intensity acoustic stimuli (80 and 90 dB in latency; 50, 60, 70, 80 and 90 dB in amplitude) (Fig. 3c). Additionally, at any acoustic intensity, the absolute values of residuals of fitting normalized latency-time curves were smaller than those of fitting the normalized amplitude-time curve, although at 40 and 50 dB, there was no statistically significant difference between two absolute values of residuals (Fig. 3d, 2 Independent Samples Tests). For latency, the absolute values of residuals of fitting at various intensity acoustic stimuli were similar to each other (Fig. 3d, the blue column). For amplitude, the absolute values of residuals of fitting at low intensity acoustic stimuli (40 and 50 dB) were smaller than those at high intensity acoustic stimuli (60, 70, 80 and 90 dB) (Fig. 3d, the cyan column).

Extracting changes in nerve fibers and synapses from AEP latencies during anesthesia

Recently, we have developed a method (DCASF) to first spike latency-intensity curves recorded in single cells, to extract changes in the theoretical minimum and stimulus-dependent components of latency57. The penetration depth was 100–1800 μm from the surface of the ICC. During penetration, the AEP was detected with a sample rate of 25 kHz by presenting 50 ms noise bursts (60 dB) to the mouse. The electrical signals were amplified 10,000 times, filtered by a band-pass of 10–300 Hz with a digital amplifier RA16 and recorded and displayed with Brain Ware software.

After AEP was detected in the ICC, a frequency scan (F-scan) was performed in which pure tone bursts (from 2 to 64 kHz in 1 kHz steps, 80 dB SPL, 50 ms duration, 5 ms rise-fall time) were randomly presented at a rate of 1/s with 10 repetitions to identify the BF of the recording site (Fig. 1c). Then, the mouse was anesthetized with sodium pentobarbital (60–70 mg/kg i.p.). Once the mouse was immobilized on the table without paw withdrawal (approximately five minutes after pentobarbital injection), an intensity scan (I-scan) was performed and was repeated every 10 min until the mouse limbs autonomously moved and paw withdrawal occurred. In this scan, the frequencies of pure tone bursts (50 ms duration with a 5 ms rise-fall time) were set at BF and were randomly presented from 30 to 90 dB SPL in 10 dB steps at a rate of 1/s and repeated 10 times. The AEPs (time window: 500 ms after stimulus onset) and the acoustic stimuli parameters in a recording session were recorded and stored in a DAM file.

The exposed brain was treated with physiological saline continuously to prevent the tissue from drying and the pinnae were maintained as in a normal awake mouse during recording. After the recording, pontamine sky blue was iontophoretically (~20 μA, 15 min) applied to the recording site with a microiontophoresis (Neurophore BH-2, Holliston, MA) to confirm the recording location in ICC. Data recorded outside the ICC were discarded.

The above AEP recording was performed without body warming, temperature monitoring and SPO2 monitoring for mice. Pentobarbital can induce hypothermia35 and hyoxemia34. And hypothermia51 and hyoxemia45 can change AEP. To analyze the effects of hypothermia and hyoxemia on AEP, the rectal temperatures (home-made thermometer) of eleven mice and the SPO2 (pulse oximetry, Nellcor, N-550) of five mice were measured every 10 min after pentobarbital injection without body warming. In addition, with body warming (home-made homeothermic blanket), rectal temperature monitoring and SPO2 monitoring, AEP was also re-recorded in five mice, as AEP recording in mice without body warming.

Ketamine, urethane and sodium pentobarbital are three common anesthetics for animal experiments. Ketamine is usually mixed with other anesthetics for use58. Urethane has a long anesthesia time59. Thus, we chose sodium pentobarbital for our experiment, as it can be used alone and has a relatively short anesthesia time (2–3 h, preliminary experimental data). In addition, noises and clicks with a wider frequency range can activate more auditory pathways60. Therefore, tones, which activate a relatively simple auditory pathway, were chosen as the acoustic stimuli to reduce data variability.

Data processing

For off-line data processing, we employed a custom-made Matlab program. First, we extracted and averaged the ten AEPs corresponding to each identical stimulus to measure the latency and amplitude. Then, we plotted the latencies and amplitudes of AEPs against anesthesia time and acoustic intensities. To fit latency-intensity curves, DCASF (for details, see Results) was used. With this fitting method, we obtained latency and intensity shifts (∆L and ∆I, respectively) of all latency-intensity curves. Then, we plotted ∆L and ∆I against anesthesia time. Next, the SPO2 and temperatures were also plotted against anesthesia time. In addition, the L80-, A80-, ∆L-, ∆I- and temperature-time curves were normalized and fit with a polynomial regression equation (polynomial order = 4). After this fitting, five groups of absolute values of residuals were obtained and compared.

The values of relevant parameters were calculated in Microsoft Excel software (version 2003). All statistical analysis was performed in SPSS statistical software (version 13). Before performing appropriate parametric statistics, datasets were first tested for normal distribution (Shapiro-Wilk test) and equal variances (Levene test). Normally distributed data were presented in figures as mean ± SD. Non-normally distributed data were presented in figures as P25, P50 , P75 and mean ± SD. For two-group comparisons, unpaired t test (Normally distributed data) or 2 Independent Samples Tests (Non-normally distributed data) was applied to test significance (two-tailed). For three or more group comparisons, one-way ANOVA (Normally distributed data) or K Independent Samples Tests (Non-normally distributed data) was applied to test significance (two-tailed). For one-way ANOVA, LSD’s test was used to further compare group means. For K Independent Samples Tests, 2 Independent Samples Tests was used to further compare any two group data. In most cases, a p-value less than 0.05 was deemed significant. However, in multiple comparison 2 Independent Samples Tests, a p-value less than 0.008 (0.05/6) was deemed significant. The data fitting and plotting were carried out in OriginPro software (version 8). The figures were arranged in Canvas software (version 9).

Additional Information

How to cite this article: Huang, B. et al. Latency of auditory evoked potential monitoring the effects of general anesthetics on nerve fibers and synapses. Sci. Rep. 5, 12730; doi: 10.1038/srep12730 (2015).