Influence of Ventilation Strategies and Anesthetic Techniques on Regional Cerebral Oximetry in the Beach Chair Position: A Prospective Interventional Study with a Randomized Comparison of Two Anesthetics

CATASTROPHIC neurological injury following anesthesia in the beach chair position has been reported in the literature¹  and is thought to be due to cerebral hypoperfusion. Anatomic abnormalities in the Circle of Willis have been rarely discovered as the etiology,²  and the majority of adverse events have occurred in healthy patients with recorded blood pressures that many anesthesiologists would consider acceptable.³ 

The measurement of regional cerebral oxygenation (rSo₂) has been employed in routine clinical practice to detect cerebral desaturation in potential low-flow states for patients undergoing cardiac⁴  and vascular surgeries.^5,6  Monitoring of rSo₂ in anesthetized patients undergoing surgery in beach chair position has been evaluated⁷  and revealed a high incidence of cerebral desaturation (rSo₂ ≥ 20% below baseline).⁸  By measuring the relative concentrations of oxyhemoglobin and deoxyhemoglobin, cerebral near-infrared spectroscopy (NIRS) provides an estimate of the balance between cerebral oxygen supply and demand within the field of view.⁹ 

Reports in conscious volunteers¹⁰  and anesthetized patients without vascular disease¹¹  demonstrate a relationship between rSo₂ and both inspired oxygen fraction (Fio₂) and end-tidal carbon dioxide (PETco₂). Observational data specifically captured during beach chair positioning suggest a relationship between PETco₂ and rSo₂,¹²  and in a recent randomized controlled trial, ventilation at a higher PETco₂ was associated with higher rSo₂ values for patients anesthetized in the beach chair position.¹³ 

Anesthetic agents have distinct effects on cerebral hemodynamics and metabolism,^14,15  which may have implications for beach chair positioning. Agents with greater preservation of cerebral blood flow (CBF) to cerebral metabolic rate for oxygen (CMRo₂) ratio may allow for greater tolerance of cerebral hypoperfusion. Cerebral oxygenation appears better preserved in the beach chair position with a combination of sevoflurane–nitrous oxide compared with propofol–remifentanil,¹⁶  yet the presence of either nitrous oxide or remifentanil could have confounded results for the primary anesthetics. In an additional study,¹⁷  rSo₂ decreased less when using desflurane compared with propofol for the first 9 min of beach chair position, but the desflurane group received thiopental for induction, which might have also confounded the results. The effect on rSo₂ due to interactions between inspired gas composition and anesthetic choice is of clinical interest, especially when comparing propofol with desflurane, the halogenated ether with the greatest potential for cerebral vasodilatation.¹⁸ 

This study tested the hypothesis that modulation of Fio₂ and PETco₂ results in significant changes in rSo₂ in patients anesthetized in the beach chair position. The influence of two anesthetic techniques with distinct effects on the cerebral vasculature was tested as a secondary outcome.

A detailed description of the study protocol (ClinicalTrials.gov No. NCT01535274) has previously been published.¹⁹  This was a prospective within-group study that incorporated a randomized comparison of two anesthetic regimens (fig. 1). The study was approved by the Institutional Review Board of the University of Michigan, Ann Arbor, and written informed consent was obtained after a detailed discussion with patients regarding risks and benefits. The primary outcome was the effect of increasing Fio₂ or PETco₂ on rSo₂ in patients anesthetized in the beach chair position. The secondary outcome was the effect of desflurane versus propofol on ventilation-related changes in rSo₂ and on cerebral desaturation in patients anesthetized in the beach chair position.

We recruited adult patients scheduled for elective arthroscopic shoulder surgery in the beach chair position under general anesthesia with a supplemental interscalene brachial plexus block at an ambulatory surgery facility. All surgeries were conducted by a single surgeon (B.S.M.); regional and general anesthetic procedures were conducted by a single anesthesiologist (A.D.). Patients were excluded if they refused to give consent, could not speak English, had contraindications for an interscalene brachial plexus block, or had a history of cardiovascular disease, cerebrovascular disease, respiratory failure, or hypertension determined by (1) medical diagnosis, (2) the surrogate of pharmacologic treatment with an antihypertensive, or (3) a blood pressure measured in the surgical preoperative clinic of greater than 140/90.

Patients underwent computer-generated randomization to receive desflurane or total intravenous anesthesia with propofol for maintenance of anesthesia. These agents were chosen because of their differential effects on cerebral vasodilation, autoregulation, and carbon dioxide responsiveness.^15,20,21  Patients and data analysts were blinded to the anesthetic choice; the anesthesiologist caring for the patient was blinded to rSo₂ values. If the absolute rSo₂ was less than 55% or decreased from baseline by greater than or equal to 20%, the anesthesiologist was informed, as instructed by our Institutional Review Board.

Baseline rSo₂ was measured in the preoperative holding area with patients sitting and breathing room air. All patients had a single-shot, ultrasound-guided, interscalene brachial plexus block performed before induction of anesthesia (20 ml of 0.5% ropivacaine). The adequacy of the block was clinically assessed using a combination of motor and sensory evaluation before proceeding to the operating room. Patients were premedicated with midazolam (0.5 to 2 mg); following preoxygenation, anesthesia was induced using fentanyl (1 to 2 μg/kg) and propofol (0.5 to 2 mg/kg). For muscle relaxation, a combination of succinylcholine and/or nondepolarizing muscle relaxants was used as was deemed clinically appropriate. The patient’s trachea was intubated, the lungs were ventilated, and general anesthesia was maintained as dictated by randomization.

The primary interventions related to Fio₂ and minute ventilation were sequentially adjusted to the following five set points:

1. Fio₂ 0.3 and PETco₂ 30 mmHg—supine position

2. Fio₂ 0.3 and PETco₂ 30 mmHg—beach chair position

3. Fio₂ 1.0 and PETco₂ 30 mmHg—beach chair position

4. Fio₂ 1.0 and PETco₂ 45 mmHg—beach chair position

5. Fio₂ 0.3 and PETco₂ 30 mmHg—beach chair position

The initial tidal volume was set at 6 to 8 cc/kg body weight and minute ventilation adjusted first by changing respiratory rate rather than manipulating tidal volume. Blood pressure was recorded by noninvasive cuff placed on the nonoperative arm. No correction factor was applied to account for the difference in vertical height between the blood pressure measurement site and the Circle of Willis. All patients were placed at 80 to 90° in the beach chair position depending on body habitus. The surrogate for depth of anesthesia was the bispectral index (BIS; Covidien, USA), which was targeted to the range of 40 to 60. Nitrous oxide could potentially confound results and was therefore avoided. The first measurement in the beach chair position (set point two) was obtained either 15 min after positioning, allowing the maximal decrease in rSo₂ to occur,⁸  or immediately if cerebral desaturation (defined as an absolute rSo₂ value < 55% or a decrease from baseline of ≥ 20%) was sustained for more than or equal to 3 min in either hemisphere. Since the change in rSo₂ is typically complete and stable within 5 min following a change in inspired gas composition,⁵  rSo₂ was recorded as a “snap shot” after a minimum of 5 min at each subsequent set point. All cerebral desaturation events (absolute rSo₂ value < 55% or a decrease from baseline of ≥ 20% sustained for ≥ 3 min) were recorded and communicated by the clinical coordinator to allow intervention as deemed appropriate by the anesthesiologist: exclude and treat hypotension, exclude and treat excessive depth of anesthesia, and move to the next ventilation set point if that included increasing Fio₂ or PETco₂. If cerebral desaturation occurred moving from set point 4 to set point 5, rSo₂ was recorded before full equilibrium had been reached and Fio₂ and/or PETco₂ were increased. Hematocrit was measured at the beginning and end of the study period (set point 1 and set point 5). Demographic and intraoperative data were retrieved from the patient’s electronic anesthetic and medical records.

Blood pressure management has been detailed in the published protocol.¹⁹  Briefly, either ephedrine (5 mg) and/or phenylephrine (50 to 100 μg) were used as intravenous bolus medications for the treatment of intraoperative hypotension. If bolus dose phenylephrine was used first, we delayed the recording of results following a trial intervention by at least 8 min in order to allow rSo₂ to normalize.²²  For patients requiring more than 400 μg of phenylephrine during a 20-min period, phenylephrine by infusion (200 μg/ml) was titrated to maintain blood pressure within 20% of preoperative mean arterial pressure (MAP).

Standard American Society of Anesthesiologists monitoring was used for all patients. The BIS Quatro electrode was placed diagonally on the patient’s left forehead. rSo₂ was measured using the INVOS 5100C monitor (Covidien, USA). Before induction of anesthesia, a single trained researcher applied optodes to either side of the forehead in conjunction with the BIS Quatro sensor as recommended by the manufacturer.

The reported mean rSo₂ is 67.1 ± 6.2% for patients placed in beach chair position.⁸  Based on a previous investigation,¹¹  we expected a 6 to 8% point difference in rSo₂ due to the planned change in Fio₂ and a 2 to 4% point difference due to the planned change in PETco₂. A total percentage point increase in excess of 10 was prespecified¹⁹  as an outcome of clinical relevance. A sample size of 24 has a power of greater than 0.8 to detect a 4 to 5% increase in the planned pairwise comparisons related to the primary intervention of ventilation strategy. The power for the comparison between the two anesthetic regimens (48 subjects) is better than 85% for a difference of 6%, which was prespecified¹⁹  as a clinically important difference for the secondary outcome.

Data were analyzed with repeated-measures ANOVA with ventilation strategy as the within-subjects factor (primary outcome) and anesthetic regimen as the between-subjects factor (secondary outcome). Residuals were assessed for normality and equal variances. A post hoc Tukey honest significant difference procedure was used to correct for all pairwise comparisons between ventilation strategies. All analyses were performed using SPSS version 21.0 (SPSS Inc., USA). A P value of less than 0.05 was considered statistically significant.

Fifty-six patients were recruited and none were withdrawn (fig. 1); all patients tolerated their procedure and recovered without complication. Twenty-eight patients were randomized to receive propofol and 28 patients were randomized to receive desflurane; secondary to the misallocation of one patient, 29 patients received propofol and 27 patients received desflurane. All patients had a functional brachial plexus block before leaving the preoperative holding area. None required supplementation with additional local anesthesia. There were no significant differences in patient characteristics or baseline data between groups (table 1). When patients were awake and breathing room air, the mean rSo₂ on the left was 71%, SD 13, and on the right was 69%, SD 11. There was no significant difference in rSo₂ between operative and nonoperative sides at any of the measurement points. There were three outliers, one in the desflurane group and two in the propofol group, who displayed mean room air rSo₂ between 36 and 40%; none of the study or clinical interventions achieved a clinically meaningful increment in rSo₂ in these patients. All patients were included for complete analysis.

While maintaining Fio₂ at 0.3 and PETco₂ at 30 mmHg, there was a decrease in rSo₂ from 68% (SD, 12) to 61% (SD, 12) (P < 0.001) following beach chair positioning when compared with stable supine patients with the same ventilation parameters for the combined data set (table 2).

The overall repeated-measures ANOVA, including the first five set points, revealed that ventilation strategy had a significant within-subjects effect on rSo₂ (P < 0.001). While maintaining PETco₂ at 30 mmHg, rSo₂ improved by 5% points when Fio₂ 1.0 was delivered compared with Fio₂ 0.3 (mean, 66%; SD, 12 vs. mean, 61%; SD, 12; P < 0.001). At 1.0 Fio₂, an additional 9% point improvement was observed at PETco₂ 45 mmHg when compared with PETco₂ 30 mmHg (mean, 75%; SD, 12 vs. mean, 66%; SD, 12; P < 0.001). A total increment of 14% points was achieved comparing Fio₂ 0.3 and PETco₂ 30 mmHg (mean, 61%; SD, 12) with Fio₂ 1.0 and PETco₂ 45 mmHg (mean, 75%; SD, 12) in the beach chair position (P < 0.001). This value is also 7% points higher than that measured in supine anesthetized patients at Fio₂ 0.3 and PETco₂ 30 mmHg (mean, 75%; SD, 12 vs. mean, 68%; SD, 12; P < 0.001). With the exception of the three outliers, all patients responded consistently to the changes in ventilation strategy with the same direction of change in rSo₂. Table 2 summarizes the rSo₂ during the trial interventions for the combined sample and for each anesthetic choice. Figure 2 illustrates the percentage change from baseline rSo₂.

There was no significant interaction effect of the anesthetic on rSo₂ as a between-subjects factor at the study intervention points. Ten patients in the propofol group and nine in the desflurane group exhibited rSo₂ less than 55% or a decrease from baseline of greater than or equal to 20% sustained for more than or equal to 3 min in either hemisphere, necessitating an early increase in Fio₂ to 1.0 during the first 15 min following beach chair positioning. There was no statistically significant difference in cerebral desaturation rate between groups.

The mean time interval between interventions 1 and 2 was 16 min (SD, 5), between 2 and 3 was 10 min (SD, 3), between 3 and 4 was 12 min (SD, 3), and between 4 and 5 was 10 min (SD, 2). The mean BIS at each study intervention point was held within two points of the lower limit of our target range (40) for the combined sample (table 3). There was no difference in BIS as a between-subjects factor when comparing the propofol and desflurane groups. There was no statistically significant change in either propofol infusion rate or end-tidal desflurane throughout the study (table 4). There was no statistically significant change in systolic or diastolic blood pressure throughout the intervention period in either the propofol or desflurane groups. There was no statistically significant difference in systolic blood pressure during the intervention period seen between the propofol and desflurane groups, but the diastolic blood pressure was consistently and significantly lower in the desflurane group when analyzed as a between-subjects factor (P = 0.001) (fig. 3). The mean difference seen in diastolic blood pressure throughout the intervention period was 11 mmHg. There was no significant variation in heart rate throughout the intervention period, but heart rate was consistently lower in the desflurane group, mean 77 (propofol) versus 65 (desflurane) beats/min (P = 0.002) (fig. 4). There was no significant difference in the number of patients receiving ephedrine as a bolus medication (8 [propofol] vs. 10 [desflurane]), the total number of doses (14 [propofol] vs. 18 [desflurane]), or the total dose (70 mg [propofol] vs. 90 mg [desflurane]), administered during the study period to each of the anesthetic groups when considered as entire groups. There was no significant difference in the number of patients receiving phenylephrine as a bolus medication (five in the propofol group and seven in the desflurane group within the intervention period). The total number of phenylephrine doses (17 vs. 31) and the total dose (1,000 vs. 1,800 mg) administered to each group throughout the study period were both significantly higher in the desflurane group (P < 0.01). No patient required phenylephrine by infusion. There was a statistically significant difference between hematocrit measured at set point 1 (median, 43; interquartile range, 38 to 46) and set point 5 (median, 41; interquartile range, 38 to 44; P = 0.001).

The misallocation carried the potential to impact the results of the secondary outcome; one patient in the propofol group was excluded from the sensitivity analysis of the remaining 55 patients (28 in the propofol group and 27 in the desflurane group). The analysis revealed no significant differences compared to the full study population. As in the complete cohort, the increase in Fio₂ and PETco₂ resulted in a significant increase in rSo₂ (P < 0.001 for comparisons of each consecutive set point), which was independent of anesthetic choice (P = 0.513).

The results of this dynamic interventional study demonstrate that simple modulation of inspired gas composition leads to an elevation in rSo₂ for patients anesthetized in the beach chair position. The magnitude (14% points) of the combined intervention not only increased rSo₂ but did so to a level surpassing that measured in the supine position. The relative increments seen with increasing Fio₂ and PETco₂ were 5 and 9% points, respectively. Except for three outliers who demonstrated low baseline cerebral saturations, the responses were consistent throughout our study population. The results are not only highly statistically significant but also have exceeded our prespecified threshold for clinical relevance.¹⁹  The data can help guide the clinical management of patients anesthetized in the beach chair position and prioritize interventions for the treatment of cerebral desaturation. Our results also demonstrate that ventilation strategy is a more powerful intervention to improve rSo₂ than anesthetic choice. In this study, there was no significant interaction effect of the two anesthetics studied as a between-subjects factor at the study intervention points. The lack of an anesthetic effect in our research is a potentially generalizable observation, given the relative effects of propofol and desflurane on CBF:CMRo₂ ratio.

It is important to note that the functional relevance of rSo₂ changes was not evaluated in this study, and the risk of cerebrovascular compromise with beach chair positioning is unclear. Overt stroke was not reported in a retrospective evaluation of more than 5,000 patients anesthetized in the sitting position. The intraoperative MAP of that population was maintained at approximately 75 mmHg or greater.²³  However, the interpretation of this value is problematic because cerebral autoregulation is attenuated (CBF becomes more dependent on systemic blood pressure) with a wide range of MAP at the lower limit of autoregulation²⁴  during anesthesia in the beach chair position. Until the lower limit of autoregulation can be routinely monitored, blood pressure control alone cannot be assumed sufficient to protect patients from neurological injury during anesthesia in the beach chair position.

The modulation of inspired gas composition has been previously shown to improve cerebral oxygenation measured by NIRS in awake subjects,¹⁰  in healthy supine anesthetized patients,¹¹  and in patients undergoing carotid endarterectomy with either general or regional anesthesia.^5,6  Increasing Fio₂ during carotid endarterectomy with regional anesthesia has also been reported to reverse neurological deficits seen with carotid cross clamp placement.²⁵  Our results are consistent with these data as well as with findings from the static comparison of two PETco₂ ranges in the beach chair position¹³ ; rSo₂ was better preserved with fewer cerebral desaturation events at the higher PETco₂ range. However, our study advances the field by demonstrating that interventions related to carbon dioxide can help reverse decreases in rSo₂ in the beach chair position, supporting a causal influence. The partial pressure of arterial carbon dioxide (Paco₂) is a direct determinant of CBF²⁶ ; the impact on cerebral autoregulation remains to be fully elucidated. The combination of hypocapnia and hypotension is particularly associated with cerebral desaturation, as measured by NIRS, during anesthesia in the beach chair position.¹³ 

The measurement of cerebral oxygenation by NIRS and its application to patients anesthetized in the beach chair position remain controversial. There is wide intersubject variability, but a change is considered to be of greater significance than baseline readings.⁹  Reductions below 50% absolute value^27,28  and reductions of greater than or equal to 20% from baseline^29–32  have been associated with cerebral ischemia. Reductions of similar magnitude have been measured in patients placed in the beach chair position using the INVOS¹²  and FORE-SITE¹³  (CASMED, USA) series of cerebral oximeters but was not detected when NIRO instruments (Hamamatsu, Japan) have been used³³  in this context. As the three outliers from our study demonstrate, patients may exhibit rSo₂ below 50% or exhibit a reduction of greater than or equal to 20% from baseline without any other correlate of cerebral ischemia. The INVOS system is susceptible to extracranial signal contamination,³⁴  but data gained during sequential clamping of the external and internal carotid artery suggest that the signal is predominantly intracranial in origin.³⁵  Near-infrared light penetrates the gray matter by a few millimeters,³⁶  and therefore, oxygenation is measured only in the superficial cortex. Cerebral oximetry values are impacted by systemic blood pressure,³⁷  sensor location,³⁸  anesthetic depth for vapor-based anesthetic techniques,³⁹  and hematocrit.³⁸  The cerebral a:v ratio is assumed constant within device algorithms, and therefore, rSo₂ can vary without a true change in cerebral oxygenation.

When making a choice between general anesthetic agents for patients with potential cerebral hypoperfusion, the balance between cerebral blood supply and oxygen demand is of logical importance. In our study, statistical significance was not achieved for any difference relating to anesthetic choice, but this potentially relates to the particular monitor used. Differences in cerebral saturation measured by an alternative methodology may have been detected. It is interesting to note that diastolic blood pressure and heart rate were consistently lower within the desflurane group and that, although a similar number of patients within both anesthetic groups required bolus dose ephedrine and phenylephrine, the total number of doses and the total dose of phenylephrine were higher in the desflurane group. Per the study protocol, measurements were avoided following phenylephrine bolus medication to avoid the potentially confounding effect on rSo₂ seen with this drug. However, differences in hemodynamic control between the two groups may have confounded the comparison between anesthetic methodologies. There are insufficient data to help estimate the potential change in rSo₂ caused by heart rate. Diastolic blood pressure shows positive correlation, approximately 1 to 2% per 10 mmHg, with cerebral oximetry in congestive heart failure.⁴⁰  It is possible that a similar relationship exists in patients without heart failure or those in the beach chair position, but this has not been determined.

There are a number of limitations to this study. The anesthesiologist was informed of cerebral desaturation when it occurred, thus allowing for protocol modification that may have impacted the results. The measured lower level of rSo₂ with beach chair positioning was potentially limited, and therefore, we may have underestimated the magnitude of the response subsequent to increasing Fio₂ and PETco₂. We waited for stability in rSo₂ following manipulation of ventilation and inspired oxygen before recording, with a mean time between measurements of 10 min or greater, but it is possible that brain equilibrium was not complete at the time of data measurement. More precise reporting of respiratory gases and blood pressure could have been facilitated by the placement of intra-arterial catheters. A small number of active smokers and patients with airway disease were included; PETco₂ may not have reliably reflected Paco₂ in these patients. However, we have demonstrated predictable increments in both Pao₂ and Paco₂ with similar inspired gas modulation in previous studies,⁶  and for healthy individuals, PETco₂ provides a reliable estimate of Paco₂.⁴¹  Invasive blood pressure measurement with the transducer zeroed at the level of the Circle of Willis would have compensated for differences in patient height and the angle of beach chair placement (which could have confounded the secondary outcome) and would have allowed better appreciation of changes in cerebral perfusion pressure. Our initial plan¹⁹  involved the measurement of hematocrit at each set point. This proved practically difficult during the first few subjects so the protocol was amended to include a hematocrit check at points 1 and 5 only. It is possible that the small statistically significant negative change in hematocrit (median, 43 to 41) during the entire study period blunted the effect of our interventions designed to improve rSo₂,³⁸  but the impact in all likelihood is clinically negligible. Unmeasured potential confounders include cerebral a:v ratio and CMRo₂, which present major challenges for data interpretation based on cerebral NIRS methodology.

In conclusion, increasing Fio₂ and PETco₂ resulted in a reliable and measurable increase in rSo₂ that overcame cerebral desaturation associated with general anesthesia in the beach chair position. Furthermore, ventilation strategy had a greater influence on rSo₂ than choice of anesthetic.

Supported by the National Center for Advancing Translational Sciences of the National Institutes of Health, Bethesda, Maryland, under Award Number UL1TR000433. The INVOS 5100C cerebral oxygenation and BIS (bispectral index) monitor were lent for this project, and cerebral oxygenation optodes were provided at no cost by the manufacturer (Covidien, USA). Additional funding was gained from the Department of Anesthesiology, University of Michigan, Ann Arbor, Michigan.

The authors declare no competing interests.

Full protocol available at: ppicton@med.umich.edu. Raw data available at: ppicton@med.umich.edu.