Loss of spectral alpha power during spine surgery: what could be wrong?

Abstract

The electroencephalographic signatures of anesthetic drugs relate to a specific set of action mechanisms within the neural circuits. During intraoperative care, the recognition and correct interpretation of the EEG spectrogram can be used as a tool to guide anesthetic administration. For example, loss of alpha power during propofol anesthesia may be a sign of lighter level of hypnosis and/or of an increase in nociceptive inputs.

We describe a case report of inadvertent interruption of propofol delivery that was first detected by changes in the electroencephalogram spectrogram.

Editor

A 51-year-old patient was presented for microsurgical resection a cervico-dorsal arterial-venous malformation. The patient received Total Intravenous General Anesthesia with tracheal intubation and mechanical ventilation. Anesthetic induction and maintenance were achieved using Target Controlled Infusion (TCI) of propofol, remifentanil and dexmedetomidine. Loss of consciousness, detected by loss of corneal reflex with saline drop and presence of slow and alpha oscillations on the electroencephalogram (EEG), occurred at effect-site concentrations of 2.5 μg/mL of propofol (modified Marsh model [1]) and 0.3 ηg/mL of dexmedetomidine (Hannivoort model [2–3]. TCI of remifentanil (Minto model [4–5]) at 2.5 ηg/mL was initiated after loss of consciousness. The EEG was monitored with a 4-channel frontal montage (F7-Fp1-Fp2-F8, reference electrode at AFZ and ground electrode at Fpz) using the SedLine^ⓡ monitor (Masimo, Irvine, CA, USA). In addition, intra-arterial pressure, motor and somatosensory evoked potentials were monitored intraoperatively.

At 160 minutes of uneventful surgery and anesthetic maintenance (stable effect-site concentrations: propofol 2.5 μg/mL, remifentanil 2.0 ηg/mL and dexmedetomidine 0.4 ηg/mL), the spectrogram showed a significant bilateral loss of power at the alpha bandwidth around 12 Hz and an increase in faster activity at 15–16 Hz (Figure 1) without any changes in the hemodynamic and vital signs. With this information, the anesthesiologist noticed that the pump running the propofol infusion had not restarted after a recent switch of the propofol syringe, and further examination showed that the alarm settings of the pump was not working properly. At the time of the incident, the calculated effect-site concentration of propofol had decrease to 2.1 μg/mL. After a closed titrated increase in the effect-site concentration of propofol, the power of alpha frequencies bandwidth returned to the normal spectrographic pattern (Figure 1). Interestingly, the Patient State Index (PSI) was always within the recommended range of surgical anesthesia (25–50). No intraoperative recall was detected after evaluation using Brice’s interview [6] at 2, 24 and 72 hours after the surgery.

Figure 1.

Spectrogram showing in fucsia boxes the interruption in bilateral stable predominant delta-alpha bandwidths by a transitory shift towards higher frequencies in slow beta frequencies ( black arrows)

Discussion

The electroencephalographic signatures of different anesthetic drugs have been described and reviewed recently [7], suggesting a neurophysiological based paradigm for intraoperative brain monitoring, where it can be used as a tool to guide anesthetic dosing [8–9]. Although color spectral analysis has some limitations [10], it is known that perception of warm colors is unique in human vision so changes in that range of wavelengths have early detection [11]. Despite the increasing number of monitors incorporating the option of spectrogram visualization, current anesthesiology training is not sufficient to identify when the hypnotic depth suggested by the EEG is discordant with the processed EEG index [12].

Propofol-based anesthesia for an adequate level of unconsciousness is characterized by the presence of incoherent slow-delta and highly coherent alpha oscillations [13–14]. Propofol anesthesia induces also two distinct patterns of phase-amplitude coupling. During induction the alpha/beta amplitudes are largest at the troughs of low-frequency oscillations - trough-max pattern; during profound unconsciousness, the phase-amplitude modulation shifts with the alpha/beta amplitudes are largest at the peaks of low-frequency oscillations - peak-max pattern. During awakening and lighter levels of anesthesia, the phase-amplitude modulated reverted to the trough-max pattern [13].

Dexmedetomidine induces dose-dependent changes in the EEG [7,15–17], increasing slow-delta oscillations across the entire scalp, theta oscillations in occipital regions, spindle oscillations in frontal regions, and decreasing beta oscillations across the entire scalp. With higher doses, the alpha-spindles disappear with an increase in the power of delta-theta oscillations. Dexmedetomidine has an anti-nociceptive effect at lower doses, inhibiting the descending nociceptive transmission by activating inhibitory interneurons that synapse onto projection neurons in the dorsal horn of the spinal cord; such low doses are not expected to induce significant and predominant changes in the EEG [7,17, 18–19].

Within a stable anesthetic state, predominantly induced by propofol, the identification of the loss of alpha-activity in the spectrogram could be the result of a decrease in propofol concentration within the plasma/effect site or an increase in nociceptive inputs [20–21]. In our case, after ruling out a sudden and sustained nociceptive surgical stimuli, the loss of alpha power could have only be due to a disruption of anesthetic delivery (intravenous access, infusion lines or infusion devices). A near-simultaneous verification of those three sources of interruption of anesthetic delivery showed that the TCI infusion pump was stopped, and it did not set off the alarm.

Niu K et al. [21] also reported a case of propofol infusion pump failure detected by the increase in the Narcotrend^ⓡ stage and index. In the present case, the proprietary index of anesthesia depth, PSI, was always within the normal range for surgical anesthesia (25–50), suggesting that the reduction in the power of the alpha bandwidth of frequencies, until the moment that the near-miss was detected, was not enough to change the index, despite the interruption of propofol delivery.

In summary, with this report we highlight the relevance of intraoperative EEG monitoring including the benefits of a monitor displaying the spectral analysis of the EEG. The changes seen in the spectrogram were caught before there was a change in the processed EEG index. Lastly, like the event of ultrasound imaging as an essential feature of modern regional anesthesia, we want to contribute to the paradigmatic evolution of a basic electroencephalographic pattern training in order to prevent intraoperative awareness and recall and/or a surgical catastrophe.