Effect of prone position on cerebral hemodynamics and noninvasive intracranial pressure assessed using transcranial Doppler in patients undergoing spine surgeries: a prospective observational study

Abstract

Background

The prone position, frequently used in spine surgeries for optimal surgical access, induces physiological changes in cardiovascular and respiratory parameters. Increased intraabdominal and intrathoracic pressures lead to elevated central venous pressure (CVP). Along with raised intrathoracic pressure, positive end-expiratory pressure (PEEP) results in elevated CVP, impeding venous blood flow from the brain and potentially affecting intracranial pressure (ICP).¹ Transcranial Doppler (TCD) ultrasound is a non-invasive method commonly used to measure cerebral hemodynamic parameters, including peak systolic velocity (PSV/ MCAv_peak), mean flow velocity (MFV/MCAv_mean), pulsatility index (PI) and resistivity index (RI), which are associated with cerebral vascular resistance, intracranial pressure, and cerebral perfusion pressure (CPP).

Method

Thirty-three patients undergoing spine surgery were assessed. The vital and TCD parameters PSV/MCAv_peak, MFV/MCAv_mean, PI, and RI were noted in the supine position. (T_a). General anaesthesia was administered, and TCD measurements were repeated after induction. (T_s). Patients were then positioned prone, and TCD measurements were repeated at intervals T_p0 - immediately after the prone position, T_p15 -15 min of the prone position, T_p30 -30 min, T_p45 -45 min and T_p60 − 60 min. Vital parameters were noted at the above-mentioned time points. Non-invasive ICP (nICP) was calculated.

Results

There was a statistically significant decrease in the heart rate (HR) compared to the supine position at T_p45 and T_p45 as compared to T_p0. There was a statistically significant decrease in systolic blood pressure (SBP) as compared to T_a at T_p0 (p < 0.001), T_p15 (p < 0.001), T_p30 (p = 0.003), T_p45 (p = 0.001), and T_p60 (p = 0.018). The study found no statistically significant changes in cerebral hemodynamic parameters (PSV/MCAv_peak, MFV/MCAv_mean, PI and RI) and nICP at various time points.

Conclusion

Our findings suggest that the prone position does not cause significant changes in cerebral hemodynamics and nICP.

Trial registration

CTRI/2023/06/053677 dated 08/06/2023.

Background

The prone position is the most commonly used position in spine surgery, offering surgeons optimal access to the operative site [1]. However, this position may result in various physiological alterations in cardiovascular, respiratory, and cerebral dynamics. The increase in intraabdominal and intrathoracic pressures during prone positioning leads to a significant rise in venous pressure, thus altering blood flow dynamics and potentially affecting circulation and organ perfusion [2, 3]. The raised intrathoracic pressure due to prone positioning causes a reduction in venous return, leading to increased central venous pressure (CVP). Elevated CVP can impede blood flow from the brain, leading to venous congestion and escalated resistance to blood flow in the cerebral vasculature, which may raise intracranial pressure (ICP) [4, 5]. In addition, the application of PEEP may also result in an increase in ICP and a consequent decrease in CPP [2, 6, 7].

Transcranial Doppler ultrasound (TCD), “the stethoscope of the brain,” is commonly utilised to monitor cerebral blood flow velocity (CBFV) in major cerebral vessels [8]. The parameters measured by TCD include mean flow velocity (MFV/MCAv_mean), peak systolic velocity (PSV/MCAv_peak), diastolic velocity, pulsatility index (PI), and resistivity index (RI). Among these, PI tends to be more often measured for its association with ICP than RI. This preference likely arises as PI relies on MFV/MCAv_mean, which considers the flow velocity throughout the cardiac cycle, whereas RI is primarily influenced by systolic velocity [8,9,10,11,12,13]. A statistically significant correlation was observed between PI and ICP in a prospective observational study by Kaloria N et al. involving 40 patients undergoing endoscopic third ventriculostomy or ventriculoperitoneal shunt surgery [14]. Various studies have been performed to determine the relationship between PI and ICP and to develop a mathematical model for nICP calculation utilising PI [10, 15].

There is a paucity of literature on the effect of the prone position on cerebral hemodynamics. Our study aims to evaluate changes in cerebral hemodynamics (flow velocity, PI, RI, nICP) in supine and prone position in thoracolumbar surgeries. We hypothesised that the prone position alters the cerebral hemodynamics. The primary objective of our study was to evaluate the effect of change in position from supine to prone on PI. The secondary objectives were to assess the impact of change in position from supine to prone on PSV/MCAv_peak, MFV/MCAv_mean, RI and nICP. We also evaluated the effects of airway pressure changes on cerebral hemodynamics after changing the position from supine to prone.

Methods

Study design

This article adheres to the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) guidelines. This single-centre, prospective observational study was conducted after the approval by the Institutional Ethics Committee (letter No. IEC/AIIMS/BTI/301; Dec 2022) and after registration with the Clinical Trial Registry of India (CTRI/2023/06/053677 dated 08/06/2023). Informed consent was obtained from all participants before inclusion in the study. Patients aged between 18 and 65 years of age, belonging to American Society of Anaesthesiologists (ASA) I and II, undergoing thoracolumbar and lumbosacral spine surgeries in the prone position in the department of Anaesthesiology in a tertiary level institute between July 2023 to March 2024 were eligible for inclusion in the study. The exclusion criteria were hemoglobin < 10 gm/dL, previous history of cranial surgery, pre-existing cerebrovascular disease, polytrauma patients, pregnancy, intramedullary spine tumours and spine trauma with dural injury.

Study sample size

The sample size was calculated for repeated measures of ANOVA within factors, taking correlation among repeated measures as 0.5, number of measurements as 6, non-sphericity correction as 0.4, eta squared was taken to detect a medium effect size = 0.06 in PI at 80% power, the sample size was found to be 33. After including 10% dropouts, the sample size was 37. The choice of the effect size was based on the theoretical expectation that should the PI in brains with preserved autoregulation vary with position, the variation should be at best a small to moderate difference from baseline and not by a large extent. This moderate change has not been documented in terms of PI with position change, and hence we chose to use the generic effect size terms (eta squared), and the cut-off of 0.06 was based on Cohen’s recommendations [16]. The sample size was calculated using Gpower [17].

Methodology

Patients were enrolled based on the aforementioned inclusion and exclusion criteria. All selected subjects were assessed preoperatively, and the study was explained in their vernacular language. Detailed patient information sheets (PIS) were provided to the patients or to their next of kin, and written informed consent was obtained before enrolment.

On the day of the surgery, patients were shifted to the operation theatre (OT) and placed in the supine position on the transport trolley. Monitors were attached from the anesthesia workstation, and non-invasive blood pressure (NIBP), electrocardiography (ECG), and pulse oximetry (SpO₂) monitoring were initiated. A wide-bore intravenous (IV) cannula was inserted in one of the upper limb veins. Bispectral Index (BIS) monitoring sensors were placed on the forehead after cleaning with an alcohol-based solution. A dedicated study investigator documented the baseline heart rate (HR), systolic blood pressure (SBP), diastolic blood pressure (DBP), and SpO₂.

TCD was performed using a 1.6 MHz probe (Delica, Model: EMS-9 PB). The middle cerebral artery (MCA) was insonated through the transtemporal window, and the best Doppler signal was achieved by fine-tuning the position of the probe, scale, gain, and angle of insonation to achieve the best signal-to-noise ratio. The point of insonation was marked with a pen, and the depth of insonation was noted for subsequent readings. Cerebral hemodynamics, including MFV/MCAv_mean, PSV/MCAv_peak, PI and RI, were assessed. All measurements were taken by a single operator on one side only, either the right or the left. Three readings were taken at one-minute intervals, and average values were noted. The first readings were designated as T awake (T_a).

Patients were premedicated with intravenous midazolam 0.01 mg/kg and intravenous glycopyrrolate 10 µg/kg. General anaesthesia was induced with intravenous fentanyl (1.5-2.5mcg/kg), titrated doses of propofol (1.5–2.5 mg/kg), and vecuronium 0.1 mg/kg. Intubation was performed using an appropriately sized cuffed endotracheal tube (7.5 mm and 8 mm for females and males, respectively). After securing the airway, general anaesthesia was maintained with desflurane in a mixture of oxygen, nitrous oxide (50:50), and vecuronium. The depth of anaesthesia was titrated to maintain a BIS value of 40–60. Patients were ventilated using volume control mode with a tidal volume of 6–8 ml/kg, PEEP of 5 cm of H₂O, an inspiratory: expiratory (I: E) ratio of 1:2, and the respiratory rate was adjusted to maintain an end-tidal carbon dioxide concentration (EtCO₂) of 30–36 mm Hg. Normothermia was maintained throughout the surgery.

After induction of anaesthesia, patients’ HR, SBP, DBP, SpO₂, EtCO₂, P peak, BIS, and MAC were noted, and assessment of MFV/MCAv_mean, PSV/MCAv_peak, PI, and RI was repeated using TCD through the same point of insonation and depth of insonation, which was used for the initial assessment and was designated as T supine (T_s). Patients were simultaneously preloaded with deficit fluid (body weight X 1.5 ml X hours fasted) to maintain normovolemia before turning the patient to the prone position. Additionally, intravenous fentanyl 0.5 mcg/kg and 1 mcg vecuronium were repeated before positioning the patient prone. 100% Oxygen was administered before disconnecting the circuit. Mean arterial pressure (MAP) was maintained at more than 60 mm Hg. Intravenous Mephentermine 6 mg boluses were used if the MAP decreased to more than 20% from baseline values.

The patient was carefully turned to the prone position on the OT table on two bolsters, one for the shoulder and one for the iliac crests. It was ensured that the abdomen was free. A head-positioning device was used to ensure a head and neck alignment in the neutral position that would avoid neck compression, provide adequate venous drainage, and reduce kinking in the endotracheal tube. The axilla and knees were cushioned appropriately, and eye padding was used to reduce the effects of pressure on the eyes. The arms were positioned at < 90° angles at the shoulder and axilla. The position of the table was adjusted so that the heart was above the level of the tragus to improve cerebral venous drainage. This position was maintained throughout the surgery.

The aforementioned measurements were repeated at specific times: immediately after putting the patient in the prone position (T_p0), after 15 min of prone position (T_p15), after 30 min of prone position (T_p30), after 45 min of prone position (T_p45), and after 60 min of prone position (T_p60). The surgery was not interrupted to perform these measurements. Vital parameters during the surgery were recorded throughout the procedure. The duration of the surgery, intraoperative blood loss, fluid intake, urine output, and intraoperative events were noted at the procedure’s completion. Non-invasive ICP was calculated using a PI value obtained through TCD using the formula ICP = 4.47 × PI + 12.68 [15] (Fig. 1).

Fig. 1

STROBE flow chart

Statistical analysis

The statistical analysis was performed using the statistical software The Jamovi project (2024) (Jamovi version 2.5). Data collected was presented as tables and graphs. Data was summarised as mean and standard deviation for continuous variables and frequency and percentage for categorical variables. A comparison of continuous variables measured at different time points was performed using Repeated Measures ANOVA. A p-value of < 0.05 was considered statistically significant. Linear mixed effects modelling was used to study the effect of gender (fixed effect) on PI, RI and ICP at different positions (random effect) accounting for repeated measures.

Results

In this study, 37 participants were enrolled after obtaining informed written consent. Four participants were excluded from the study due to inadequate transtemporal window visualisation (Fig. 1). The demographic parameters and indications for surgery have been mentioned in Table 1.

Table 1 Demographic parameters

Among the cerebral hemodynamic parameters and nICP, there were no significant differences in PSV/MCAv_peak (p = 0.423), MFV/MCAv_mean (p = 0.353), PI (p = 0.719), RI (p = 0.843) and nICP (p = 0.719) across the various time intervals and positions studied (Table 2; Fig. 2).

Table 2 Hemodynamic, ventilatory, anesthetic and transcranial doppler parameters at different time points

Fig. 2

Peak systolic velocity (PSV), Mean flow velocity (MFV), Pulsatility index (PI), Resistivity index (RI) and Noninvasive Intracranial pressure (nICP) at different time points. Boxplot of TCD parameters: A: PSV, Peak systolic velocity; B: MFV, Mean flow velocity; C: PI, pulsatility index; D: RI, resistivity index; E: nICP, non-invasive Intracranial pressure at various time points. (Black squares represent mean values, Grey dots are individual data points)

The HR was recorded at different positions and time intervals among the hemodynamic parameters. There were significant differences in HR across the various time intervals and positions studied (p < 0.001) (Table 2; Fig. 3A). A post hoc test analysis was done, and HR at T_a was compared with HR at T_s, T_p0, T_p15, T_p30, T_p45, and T_p60. There was a statistically significant decrease in HR with a mean difference of 7 ± 2 beats/min at T_p45. There was no statistically significant difference when HR at Ts was compared with HR at T_p0, T_p15, T_p30, T_p45, and T_p60. When HR at T_p0 was compared with HR at T_p15, T_p30, T_p45, and T_p60., HR at T_p45 showed a statistically significant decrease in HR with a mean difference of 6 ± 1 beats/min (Supplemental digital content 1; table showing post hoc analysis of HR at different time points). There was a significant difference in SBP across the various time intervals and positions studied. (p < 0.001) (Table 2; Fig. 3B). During the post hoc test analysis, there was no statistically significant decrease in SBP at T’_s, while at T_p0, T_p15, T_p30, T_p45 and T_p60, a statistically significant decrease in SBP was observed compared to T_a with a maximum mean difference at T_p45 (12.606 ± 2.70). (Supplemental digital content 2; table showing posthoc analysis of SBP at different time points)

Fig. 3

Heart rate (HR), Systolic blood pressure (SBP) and Diastolic blood pressure (DBP) at different time points. Boxplot of hemodynamic parameters: A: HR, Heart rate; B: SBP, Systolic blood pressure; C: DBP, Diastolic blood pressure at various time points. (Black squares represent mean values, Grey dots are individual data points)

There was a significant difference in DBP across the various time intervals and positions studied. (p < 0.001) (Table 2; Fig. 3C) On post hoc test analysis, no statistically significant difference was observed at T_a, while at T_p0, T_p15, T_p30, T_p45 and T_p60, a statistically significant decrease in DBP was observed with a maximum mean difference at T_p0 (10.6364 ± 1.78) (Supplemental digital content 3; table showing post hoc analysis of DBP at different time points).

There were no significant differences in EtCO₂ across the various time intervals and positions studied. (p = 0.672) (Table 2). The peak airway pressure (P_peak) was recorded, and there were no significant differences in P_peak across the various time intervals and positions studied. (p = 0.5498) (Table 2) There were no significant differences in minimum alveolar concentration (MAC) values across the various time intervals and positions studied. (p = 0.3535) (Table 2). There were no significant differences in Bispectral index (BIS) scores at BIS- T_p0, BIS- T_p15, BIS- T_p30, BIS- T_p45, BIS- T_p60 compared to BIS-T_s. (p = 0.9646) (Table 2) There was no statistically significant change in the PI, RI or ICP with respect to gender (evidenced by the CIs crossing the line of no difference at 0 in the supplemental digital content 4).

Discussion

We conducted a prospective observational study on the effect of the prone position on cerebral hemodynamics as assessed using TCD in patients undergoing thoracolumbar spine surgeries. The primary objective of our study was to evaluate the effect of change in position from supine to prone on PI. The secondary objectives of our study were to assess the change in CBFV, RI and the effect of the prone position on nICP.

Thirty-seven patients satisfied the inclusion and exclusion criteria. Among these, four patients were dropouts due to an inadequate temporal window. Thus, the data of 33 participants was analysed. The mean age of the participants was 40.5 ± 11.4 years. There were 72.7% males and 27.3% females in our study.

There was no statistically significant difference in parameters of cerebral hemodynamics, including PSV/MCAv_peak (p = 0.423), MFV/MCAv_mean (p = 0.353), PI (p = 0.719) and RI (p = 0.843). Bombardieri AM et al. also found no statistically significant difference in cerebral blood flow CBFV_syst (the maximal CBFV during the systolic phase of a cardiac cycle) and CBFV_mean (the time-averaged value of the maximal velocity envelope over one cardiac cycle) when the patient’s position changed from supine to prone position and at different time intervals after prone position. There was no statistically significant difference in PI immediately after the prone position, 15 min after the prone position, 30 min after the prone position and 45 min after the prone position. However, there was a statistically significant increase in PI observed 60 min after the prone position (p < 0.05) in the study by Bombardieri AM et al. They attributed this increase in PI to the older mean age (mean age = 58 ± 13 years) of patients in their study. They stated that the cerebral arteries may become more rigid with age, leading to higher pulsatile flow and PI [18]. Our study had a relatively younger population (mean age = 40 ± 10 years), which might explain the absence of a statistically significant change in PI at various time points in our patients.

We found no statistically significant difference in nICP between supine and prone positions at various time points. We also observed that the addition of PEEP (5 cm H₂O) did not alter PI and nICP in our study. On the contrary, a study by Robba C et al. estimated the change in ICP when the patient was in the prone position by TCD analysis of PI (ICP_PI = 4.47·PI + 12.68 mm Hg) and found a significant increase in the mean values of ICP_PI after a change from supine position to prone position in patients undergoing spine surgeries. They also found a further increase in ICP_PI after applying PEEP (8 cm H₂O) in the prone position [6]. This increase in ICP_PI could be due to the patients’ older mean age (mean age = 54 ± 16.4 yrs) in their study. Secondly, they applied a higher PEEP in their patients (PEEP of 8 cmH₂O) than in our study, wherein a physiological PEEP of 5 cm of H₂O was used [6]. This higher PEEP could have led to increased ICP through an increase in intrathoracic pressure and CVP, resulting in reduced cerebral venous return. Secondly, increased PEEP can increase spinal pressure, reduce cerebrospinal outflow, and increase ICP. Thirdly, a higher PEEP can cause a reduction in MAP, leading to cerebral vasodilation and increased ICP. We found no statistically significant difference in nICP, which could be attributed to the relatively younger mean age in our study (40 ± 10 yrs) and consistent application of physiological PEEP (5 cm H₂O) throughout the procedure. The changes in ICP primarily depend on cerebral autoregulation, defined as the ability of the cerebral vasculature to maintain constant cerebral blood flow despite the change in systemic blood pressure [19, 20].

In our study, the hemodynamic parameters (HR, SBP, DBP) during awake, supine under anaesthesia and in the prone position at various time intervals showed statistically significant differences with a p-value of HR (p < 0.001), SBP (p < 0.001) and DBP (p < 0.001). On Post Hoc analysis, the statistically significant decrease in the HR was observed at T_p45 compared to the baseline, i.e. T_a (mean difference = 7 ± 2 bpm) and at T_p45 compared to T_p0 (mean difference = 6 ± 1 bpm.); the decrease in HR was around 10%, hence not clinically significant. The statistically significant decrease in SBP as compared to baseline, i.e. T_a was observed at T_p0 (p < 0.001). However, this difference in SBP was not clinically significant (mean difference < 20% from baseline). A statistically significant decrease in DBP as compared to baseline at T_a was seen in all the time points (< 0.001) after prone positioning and had no major impact clinically, as the difference was < 20% from the baseline. Bombardieri AM et al. found a statistically significant decrease in MAP after 60 min of the prone position (70 ± 9 mm Hg) as compared to baseline i: e before the prone position (80 ± 18 mmHg) (p < 0.05).[18,] but it was < 20% from baseline and hence was not clinically significant [18]. Another study by Robba C et al. demonstrated no statistically significant difference in SBP when a patient’s position was changed from supine to prone [6]. This lack of clinically significant change in haemodynamic parameters in our study, as in the study by Bombardieri AM et al., can be explained by the administration of an adequate amount of fluids before turning the patient prone. It was then followed by administration of an adequate amount of maintenance fluids.

We used the volume control mode of ventilation in all the patients. A PEEP of 5 cm H₂O was applied. There was no significant change in peak airway pressure (P_peak) between the supine position (T_s) and the prone position (T_p0, T_p15, T_p30, T_p45 and T_p60) (p = 0.549). This was achieved by adequately placing bolsters and ensuring the abdomen was free. A straight head and neck alignment was maintained to avoid neck compression and kinking of the endotracheal tube. There was no significant change in EtCO₂ between the supine position (T_s) and the prone position (p = 0.672). There was no significant change in MAC between the supine position (T_s) and prone position (p = 0.3535). There was no significant change in BIS scores between the supine position (T_s) and prone position.

In our study, we maintained adequate oxygenation (PaO₂) and ventilation (tidal volume 6-8 ml/kg) (EtCO₂) by using VCV mode of ventilation. There was no significant change in EtCO₂ and P_peak when the position changed from supine to prone. The haemodynamic parameters, i.e. HR and MAP, were maintained within 20% of baseline using adequate fluid resuscitation and the use of anaesthetic agents titrated to BIS. Depth of anesthesia was comparable in both supine and prone positions after the patient was anesthetised. Normothermia was maintained throughout the surgical procedure. These factors might have helped to maintain cerebral hemodynamics in our patients. These findings also suggest that in patients undergoing anesthesia for surgeries in the prone position, the hemodynamic parameters, cerebral hemodynamics, and nICP can be maintained if adequate fluid balance, oxygenation and ventilation (target EtCO₂ of 30–36 mm Hg) are maintained.

The limitation of our study was the small sample size, which could have affected its generalisability. Secondly, we did not follow the patients throughout the surgery, so we could not evaluate the hemodynamic changes that can possibly happen with longer surgical times and their effect on cerebral hemodynamic parameters (PSV/MCAv_peak, MFV/MCAv_mean, PI, and RI). Thirdly, we did not include ASA 3, 4, or 5 patients where a change in position could have significant changes. Fourthly, we used TCD and measured CBFV, which is only a surrogate marker of cerebral blood flow.

Conclusion

Our findings suggest that the prone position does not cause significant changes in cerebral hemodynamics, provided adequate fluid resuscitation before turning the patient to the prone position, and adequate oxygenation (PaO2) and ventilation (PaCO2) are maintained.