The standard-of-care guidelines published by the American Society of Anesthesiologists (ASA) recommend monitoring of pulse oximetry, blood pressure, heart rate, and end tidal CO2 during the use of anesthesia and sedation. This information can help to identify adverse events that may occur during procedures. However, these parameters are not specific to the effects of anesthetics or sedatives, and therefore they offer little, to no, real time information regarding the effects of those agents and do not give the clinician the lead-time necessary to prevent patient “awareness.” Since no “gold-standard” method is available to continuously, reliably, and effectively monitor the effects of sedatives and anesthetics, such a method is greatly needed. Investigation of the use of functional near-infrared spectroscopy (fNIRS) as a method for anesthesia or sedation monitoring and for the assessment of the effects of various anesthetic drugs on cerebral oxygenation has started to be conducted. The objective of this paper is to provide a thorough review of the currently available published scientific studies regarding the use of fNIRS in the fields of anesthesia and sedation monitoring, comment on their findings, and discuss the future work required for the translation of this technology to the clinical setting.

1. Introduction

In the United States, awareness with recall during general anesthesia has been estimated to occur in 1-2 cases per 1000 patients annually, which amounts to approximately 26,000 cases per year [13]. The classifications of events that constitute awareness are somewhat controversial and can range from feelings of pain to recall of conversations during surgery. In other countries, such as China, the incidence of anesthesia awareness has been found to be as much as two times higher than that of the United States [4], while a British audit found the incidence to be lower [5]. The studies conducted thus far have linked awareness with recall to age, female sex, obesity, anesthetist’s experience, length of surgery, and previous awareness events. Patients who have suffered from intraoperative awareness have been known to experience long term psychological consequences, most notably posttraumatic stress disorder [6].

Anesthetic agents used during general anesthesia disrupt the activity of neurons in a dose dependent manner in order to suppress memory formation and awareness. This decrease in neural activity can be visualized with positron emission tomography (PET) as a decrease of the glucose metabolic rate (GMR) and the cerebral metabolic rate of oxygen (CMRO2) consumption [79]. Furthermore, it is to be noted that all volatile anesthetics suppress the cerebral metabolic rate and affect the cerebral blood flow in a dose dependent manner [10]. This information is very important, due to the fact that the cortical regions of the brain, including the prefrontal cortex, experience functional inhibition during anesthesia [1114].

Under the current standard of care published by the ASA, continuous monitoring of oxygenation, ventilation, circulation, and temperature is required [15]. These measurements allow the anesthesia provider to detect adverse events and improve patient safety during surgery; however, they are not specific to the effect of the anesthetics on the brain. The lack of a “gold-standard” method for assessing the effects of anesthetics and sedatives on the brain has led to the investigation of electroencephalographic (EEG) signals, auditory evoked potential (AEP), and facial electromyography (EMG) signals as possible monitors of depth of anesthesia. Currently, devices based on these technologies, such as the Bispectral Index (BIS) monitor (Medtronic, Dublin, Ireland), are clinically available. The BIS monitor evaluates the Fourier domain of electroencephalograph epochs and through a proprietary algorithm outputs a number between 0 and 100, with values under 60 representing the anesthetized state. Still, although accessible, the routine use of EEG monitors has been estimated to be approximately 1.8% [16]. The low rate of use of these devices may be influenced by reduced performance of the devices in the presence of opioids [17] and the fact that they have been found to provide information that correlates with the patient’s state but are not predictive enough to guide the delivery of anesthetics [18, 19]. In contrast to other means of monitoring depth of anesthesia, near-infrared spectroscopy (NIRS) monitors change in circulatory oxygenation of the cerebral cortex, which can reflect tissue oxygen use [20, 21]. These facts guided the reasoning for investigating the sensitivity of hemodynamic response measured by functional near-infrared spectroscopy (fNIRS) on the prefrontal cortex (PFC) to determine changes in the anesthetic depth.

fNIRS is a safe, noninvasive, and portable optical method that can be used to monitor activity within the cortical areas of the human brain. Making use of specific wavelengths of light, fNIRS provides measurements of oxygenated (HbO2) and deoxygenated (Hb) hemoglobin that are in direct relation with hemodynamic changes in the brain [2022]. A close link between hemodynamics and neural activity has been established through brain-energy metabolism research [2325]. Previous studies comparing fNIRS to established neuroimaging technologies, such as functional magnetic resonance imaging (fMRI) and PET, have shown that fNIRS provides comparable results in various domains including attention, memory, and sensory/motor areas in healthy and diseased/disabled populations of various age groups [2630].

fNIRS has also been used to detect cerebral hypoxia during carotid endarterectomy [31, 32] and to monitor cerebral perfusion during liver transplantation [33]. Its capacity to determine the hemodynamic changes that occur in the cerebral cortex as a consequence of anesthetic use makes this technology an ideal candidate for the study and development of monitors for anesthetic depth. Furthermore, because it is portable and noninvasive, fNIRS can be of substantial practical use in clinical settings.

The primary purpose of this paper is to provide a review of studies of anesthesia monitoring using near-infrared spectroscopy to provide a better understanding of the relationship between measurable hemodynamic changes in the prefrontal cortex and changes in anesthetic state. This review also presents information on the challenges associated with the translation of this technology into a clinical environment. This review will begin by outlining the results of investigations of anesthetic drug effects on cerebral metabolism and hemodynamics, continue with studies that evidence potential confounding factors for the fNIRS signal, and conclude with studies that have uncovered potential biomarkers in the fNIRS signal that could signal changes in anesthetic depth.

2. Methods

The study selection was performed using the PubMed database to find studies that investigated brain hemodynamics during anesthesia or sedation using functional near-infrared spectroscopy. The search was restricted to English language articles published between January 1, 1990, and May 1, 2015. The search terms included “functional near infrared spectroscopy,” “fNIR,” “NIRS,” “anesthesia,” “sedation,” and “neuromarker.” During the search, journal articles with topics related to oxygen saturation monitoring during cardiac or other surgeries were excluded. The results of the studies were divided into three groups: (i) findings on the effects of anesthetics on cerebral hemodynamics, (ii) findings related to confounding factors of the fNIRS signal, and (iii) findings on fNIRS biomarkers of depth of anesthesia. Tables 13 contain a summary of the findings of the studies presented in this paper.

To further inform the reader on the topics examined in this review we present representative example plots from our data obtained during a study of 50 patients undergoing elective surgery at the Hahnemann University Hospital, Philadelphia, PA. The study protocol and statements of informed consent were approved by the Institutional Review Board (IRB) of Drexel University. The study was conducted with the subjects’ understanding and consent. The figures presented in this review originated from fNIRS data recorded during three elective surgical procedures, of patients classified as ASA class I or II, under general anesthesia using propofol for induction and sevoflurane for maintenance. In this review, data from this study was used to serve as examples, when relevant, for the results described in previous studies.

3. Results

The results of the literature search were grouped by the following topics: (i) findings on the effects of anesthetics on cerebral hemodynamics, (ii) findings related to confounding factors of the fNIRS signal, and (iii) findings on fNIRS biomarkers of depth of anesthesia.

3.1. Effects of Anesthetics on Cerebral Hemodynamics

Initial studies of fNIRS in the area of anesthesia examined primarily the differences in cerebral metabolism and hemodynamics elicited by different drugs on the prefrontal cortex. These investigations focused on examining cerebral oxygen saturation (rSO2), cerebral blood volume as approximated by the total hemoglobin concentration (HbTot), and cerebral blood flow (CBF) in response to the administration of anesthetics. For clarity, this review separates the studies into two categories: (i) evaluation of fNIRS data during induction and (ii) evaluation of fNIRS data during steady state anesthesia maintenance. A summary of the findings of studies included in this section can be found in Table 1.

3.1.1. Effect of Anesthesia Induction

Anesthetic drugs are known to decrease the CMRO2 and to have a drug and dose dependent effect on CBF. The examination of fNIRS derived hemodynamic changes during the induction of anesthesia provides a method for examining the effects of different drugs during the transition between the awake and anesthetized state.

Using fNIRS, Lovell et al. documented small changes in cortical brain oxygenation during the induction of anesthesia with propofol, thiopental, or etomidate by following the adult patient’s hemodynamic changes from a baseline period until 3 minutes after the administration of the intravenous (IV) drugs [34]. Changes in oxygenation were reported as differences in the relative HbO2 and HbTot concentrations. This study found that propofol and thiopental induce an increase in both HbO2 and HbTot relative to the baseline period [34]. In our 50-patient study, we have observed similar changes in HbO2 and HbTot following propofol induction. Figure 1 shows an example of the changes in oxygenation that occur after induction of anesthesia with propofol. In contrast, etomidate exhibited a decrease in both HbO2 and HbTot parameters, possibly due to the different effect of this drug on oxygen saturation and blood flow [34]. This study revealed the capacity of fNIRS to measure drug dependent changes in oxygenation associated with anesthetic induction. A similar study on a pediatric population examined the effects of propofol induction on cerebral oxygenation of the frontal cortex [35]. As in the study by Fleck et al., propofol was found to produce an increase in HbO2 and HbTot relative to the conscious period [35].

Iwasaki et al. examined the changes in HbO2 and HbTot that occur after induction with 67% N2O and propofol, sevoflurane at 5%, or sevoflurane at 8% [36]. The patient’s hemodynamic changes were continuously recorded from preinduction period until three minutes after intubation. This study reported significant increases in HbO2 and HbTot after 8% sevoflurane induction compared to 5% sevoflurane and propofol [36]. This suggests that fNIRS is sensitive to changes caused by different concentrations of the same anesthetic. This is evidenced by the difference seen between 5% and 8% sevoflurane. However, this study does not match the findings of the other propofol induction studies outlined above. The results could be explained by the use of N2O in conjunction with propofol during the induction procedure, which differs from the method employed in the studies by Lovell et al. and Fleck et al.

These initial studies provided the ground work to demonstrate that fNIRS is capable of monitoring real time changes that occur as a consequence of the administration of anesthetics. Additionally, it was shown that this technology is capable of distinguishing between the effects of induction with different drugs, such as propofol, etomidate, and thiopental [34]. Different anesthetic concentrations of the same drug, such as 5% versus 8% sevoflurane, also registered different fNIRS signal signatures during induction [36].

3.1.2. Evaluations during Anesthesia Maintenance

Initial studies of fNIRS parameters during maintenance focused on examining differences in rSO2 and CBF between anesthetics to understand the effect of anesthetics on the oxygen supply demand balance and metabolism. Differences in oxygen saturation on the frontal cortex were found between sevoflurane and propofol during maintenance of anesthesia [37]. Sevoflurane was found to produce higher rSO2 values, which indicated that sevoflurane at 1.5% maintains the CBF/CMRO2 coupling to a similar extent as propofol [37]. Measurements of rSO2 have also been used to evaluate the incidence of cerebral desaturation events during surgical procedures [37, 41]. An investigation of such events in adults between the ages of 18 and 65 during propofol and sevoflurane anesthesia found that desaturation occurred only during the use of propofol [37]. On the other hand, during a sevoflurane study of elderly patients cerebral desaturation occurred at 26% of the examined cases [41].

A study by Fassoulaki et al. [38] examined differences on rSO2 between sevoflurane and desflurane during maintenance. First, rSO2 was examined during maintenance with BIS values between 40 and 50. No difference in oxygen saturation was found between the inhalants desflurane and sevoflurane [38]. Next, rSO2 was measured during maintenance of BIS values between 20 and 30 by increasing the concentrations of the inhalants. Again, no significant difference in rSO2 between inhalants was found [38]. However, this study found that rSO2 was higher when BIS values were maintained between 20 and 30 than the ones obtained when BIS values were kept between 40 and 50 [38]. During the 15 minutes of the measurement at steady state maintenance, the rSO2 level was stable [38]. The study by Fassoulaki et al. [38] provided evidence that rSO2 changes with increasing concentrations of sevoflurane and desflurane.

Kanemaru et al. [39] examined the effect of administration of midazolam, isoflurane, or aminophylline during maintenance of anesthesia with propofol. The goal of this study was to determine if the BIS index changes that occur with additional drug infusion are mirrored by changes in oxygenation. It was found that the three previously mentioned drugs affected the BIS index but not rSO2, suggesting that they may only have a negligible effect on oxygenation [39].

fNIRS has also been employed to examine cerebral blood flow during anesthesia maintenance. A study quantifying CBF during steady state anesthesia maintenance with isoflurane versus conscious patients showed no difference between the two states [40].

The previously outlined studies offer evidence for the capacity of fNIRS to measure differences in oxygenation between types and concentrations of anesthetics that could help in the assessment of anesthetic depth [37, 38]. Moreover, the studies demonstrated that some anesthetic drugs (midazolam, aminophylline, and isoflurane) may affect anesthetic depth, while having no effect on oxygen saturation [39]. To determine the effects of midazolam, aminophylline, and isoflurane on brain oxygenation an investigation of all the fNIRS derived measures (i.e., Hb, HbO2, and HbTot) is necessary. Additionally, the studies indicate that oxygenation remains constant when the depth of anesthesia is maintained at a steady state [38].

3.2. Findings on Confounding Factors of the fNIRS Signal

Several factors can affect the oxygen supply demand balance during fNIRS measurements of patients under anesthesia, affecting the interpretation of the hemodynamic variables recorded by the device. It is important to understand how the fNIRS signal can be affected by these confounding factors so that algorithms can be developed to suppress the effect of confounding information. These factors can be related to the position, type of surgery, hematocrit levels, and ventilation. A summary of the findings of studies included in this section can be found in Table 2.

3.2.1. Effects of Position

Position change, such as various tilt angles of the operating table, has been demonstrated to affect the hemodynamic parameters measured by fNIRS [4246]. The hemodynamic changes observed after position change may be related to changes in CBV associated with the effect of gravity but are also dependent on the anesthetic. Changing to the Trendelenburg position after CO2 pneumoperitoneum was evaluated during propofol and sevoflurane anesthesia. Significant increase in rSO2 after CO2 insufflation and position change was present with the use of sevoflurane but not propofol [42]. This study also found lower rSO2 values for propofol anesthesia [42].

Changing to the sitting position was evaluated during anesthesia maintenance with propofol and desflurane anesthesia [43]. Upon shifting to the sitting position, both groups experienced a significant decrease in rSO2. The propofol group had significantly lower rSO2 values [43]. Oxygen saturation was measured during nine minutes in the sitting position and during this time no significant differences were observed within the groups. Returning to supine position restored rSO2 values to the baseline level [43]. The results show that moving to the sitting position caused a decrease in cerebral oxygen saturation in equipotent concentrations of propofol and desflurane groups, with propofol experiencing a larger decrease [43]. Similarly, movement from supine to beach chair position (BCP) was associated with a 10% decrease in rSO2 that was reversible after returning to the supine position [44]. Head-up position was also evaluated by Kitajima et al. [45] during maintenance with sevoflurane combined with 66% N2O. After changing to the head-up position, HbO2 decreased significantly and continued to decrease for 30 minutes after the movement [45]. The recorded HbO2 values remained lower than the baseline even after returning to supine position [45]. A similar effect was observed in HbTot [45]. In contrast, Hb did not experience significant changes with the movement [45].

Lovell et al. [46] examined changes in blood volume with changes in position during propofol anesthesia. These investigators found increases in CBV during the head-down position that correlated with the degree of table tilt. The degree of table tilt for the head-up position was also correlated with table tilt angle, but the changes during head-up position were markedly lower than the equivalent degree of change in the head-down position [46].

The studies outlined in this section demonstrate that changes in position can have significant effect on fNIRS derived measures such as rSO2, HbTot, HbO2, and Hb. The changes that occur depend on the type of position and the anesthetics in use. We have observed such positional effects on hemodynamic parameters as measured by fNIRS in our prior study. Figure 2(a) presents an example of the change that is observed on the hemodynamic parameters as a result of a transition from supine to Trendelenburg position. Figure 2(b) shows an example of a return to supine position from Trendelenburg position.

3.2.2. Effects of Other Confounders

The partial pressure of carbon dioxide (PaCO2) dissolved in the blood can affect the cerebral blood volume. Low levels of carbon dioxide (CO2) will cause a reduction in CBF. Alterations in intrathoracic pressure, such as during CO2 insufflation for laparoscopic procedures, can cause alterations in cerebral perfusion and PaCO2. A study by Owen-Reece et al. [47] showed that CBV, as measured by fNIRS, falls when the intrathoracic pressure is increased. Hyperventilation to decrease PaCO2 also led to a significant decrease in rSO2 in groups of propofol-remifentanil and sevoflurane anesthesia [48]. The evaluation by Kim et al. [49] focusing on the effect of preoxygenation with 100% oxygen found that administration of propofol with sufentanil or midazolam after preoxygenation did not have an effect on rSO2 values.

Another confounding element influencing the fNIRS signal is hemodilution experienced during blood loss, when the lost volume is replaced by crystalloid leading to a decrease in the volume of erythrocytes. This effect was examined via measurements of rSO2 during propofol and sevoflurane anesthesia [50]. During the study, oxygen saturation decreases were measured with both anesthetics [50]. Because fNIRS values are dependent on hemoglobin concentration, hemodilution during surgery can influence the hemodynamic parameters measured by fNIRS.

The effects of changes in mean arterial pressure (MAP) have also been evaluated as potential confounders of the fNIRS signal. Nissen et al. [51] evaluated rSO2 during blood pressure decreases that occur after IV administration of propofol and fentanyl during induction. This study found that cerebral oxygenation increased and remained stable during the surgery; however no correlation between MAP and rSO2 could be established [51]. The elevation in rSO2 during induction does however support the trends reported by other authors [34, 35].

3.3. Findings on fNIRS Biomarkers of Depth of Anesthesia

Initial fNIRS studies examined the hemodynamic effects of anesthetic drugs on rSO2, CBV, and CBF. More recent studies have focused on finding differences in the HbO2, Hb, and HbTot biomarkers that could signal transitions between anesthetic states. A summary of the findings of studies included in this section can be found in Table 3. The first study to examine fNIRS biomarkers for depth of anesthesia was performed to validate the capacity of fNIRS to discriminate between “deep” and “light” anesthesia [52]. fNIRS data was collected in the operating room during abdominal or lower body surgical procedures. Deep anesthesia was classified as the 4-minute period prior to wound closure, while light anesthesia included the 4 minutes prior to eye opening. The evaluation consisted of a comparison between the levels of HbO2, Hb, and HbTot in 26 patients undergoing surgery with general anesthesia using either inhaled sevoflurane or desflurane [52]. The investigators established that across all prefrontal cortex PFC channels there was a statistically significant decrease in the levels of Hb during the transition from deep to light anesthesia [52]. The most significant changes were found on the right side of the PFC [52]. Similar changes were also observed in our prior study. One example of the changes in the fNIRS signal observed during the transition between maintenance and emergence is presented in Figure 3.

An additional study with a sample size of 20 patients undergoing coloproctology surgery also examined the concentration changes of Hb in the prefrontal cortex during different stages of anesthesia [53]. In the cases evaluated within this study, propofol was administered to induce anesthesia and sevoflurane was the maintenance agent. The investigators found that the average Hb concentration was significantly increased from baseline after propofol induction [53]. The comparison of average Hb concentration between deep and light anesthesia showed a decrease of Hb as the patient began to wake, and this change reached statistical significance on the right side of the PFC [53].

In a similar manner, fNIRS was used to evaluate the effects of propofol sedation during 41 outpatient elective colonoscopy procedures [54]. Due to previous results, channels located on the right PFC were selected as the PFC areas of interest and the concentrations of HbO2 and Hb were analyzed before and after propofol bolus infusions at different concentrations. In this study, a dose dependent increase in HbO2 became significant at the two- and three-minute marks after propofol administration [54]. Significant positive correlation between the change in HbO2 and the administered dose was also found [54]. No significant changes were found in Hb for propofol sedation doses. These results are in accordance with previous reports showing significant increases in HbO2 following propofol induced anesthesia [34].

4. Conclusions and Future Directions

Findings evidencing a reduction of the cerebral metabolic rate of oxygen as a consequence of anesthetic use in regions that included the prefrontal cortex served as motivation for the study of the sensitivity of the hemodynamic response, as measured by fNIRS, as a monitor of depth of anesthesia. Early work investigated metabolic and hemodynamic differences between anesthetics, indicating that fNIRS was capable of measuring the small changes in hemoglobin concentration that occurred in response to different anesthetics and concentrations. Although the literature on the subject is limited, fNIRS has been shown to provide information on cerebral oxygenation that is discriminative between anesthetic states [3436, 38, 5254], concentrations of the same anesthetic [36, 38, 5254], and different anesthetics [34, 36].

Because the hemodynamic response measured by fNIRS is only capable of providing information on the hemoglobin species concentration it is an indirect assessment of the metabolic rate. As such, it can be affected by confounders that influence the relative quantities of Hb and HbO2 without changes in the metabolic rate. Position [4246], systemic CO2 and O2 concentration [4749], and hemodilution [50] are some of the possible confounders. Careful examination is needed to determine the hemodynamic relationships that are least affected by confounding factors or algorithms will need to be developed to mitigate their effects. Some investigators have already begun to devise algorithms to rule out possible confounders on the NIRS signal during routine oxygen saturation monitoring [55]; however, these algorithms have not yet been integrated into the monitoring software.

In the available literature only three studies were found that specifically investigated biomarkers for depth of anesthesia using the fNIRS signal. The studies by Izzetoglu et al. [52] and Leon-Dominguez et al. [53] found significant changes in Hb as the subject transitioned between deep and light anesthetic states during the use of sevoflurane and desflurane. The study by Curtin et al. [54] found that propofol prompted a dose dependent increase of HbO2. These findings suggest that fNIRS measured biomarkers such as Hb and HbO2 concentrations on the prefrontal cortex are sensitive to the effects of different anesthetic types and concentrations [5254]. In addition, fNIRS is sensitive to different anesthetics such as propofol, sevoflurane, and desflurane at sedation and surgical doses [5254]. Further research is needed to determine if other features of the fNIRS signal contain reliable information on the anesthetic depth. Future efforts should focus on the development of real time examination methods to assess the capacity of the fNIRS signal to provide reliable information that can be used to guide anesthesia delivery. Altogether, with its portability, affordability, safety, ease of use, and the current findings suggesting that fNIRS is indeed suitable for measuring the effects of anesthetics on the suppression of awareness and its return, this makes fNIRS attractive for further research in the field of depth of anesthesia monitoring.

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.