Evaluation of cerebral hemodynamics by transcranial Doppler ultrasonography and its correlation with intracranial pressure in an animal model of intracranial hypertension

ABSTRACT Background: Transcranial Doppler has been tested in the evaluation of cerebral hemodynamics as a non-invasive assessment of intracranial pressure (ICP), but there is controversy in the literature about its actual benefit and usefulness in this situation. Objective: To investigate cerebral blood flow assessed by Doppler technique and correlate with the variations of the ICP in the acute phase of intracranial hypertension in an animal model. Methods: An experimental animal model of intracranial hypertension was used. The experiment consisted of two groups of animals in which intracranial balloons were implanted and inflated with 4 mL (A) and 7 mL (B) for controlled simulation of different volumes of hematoma. The values of ICP and Doppler parameters (systolic [FVs], diastolic [FVd], and mean [FVm] cerebral blood flow velocities and pulsatility index [PI]) were collected during the entire procedure (before and during hematoma simulations and venous hypertonic saline infusion intervention). Comparisons between Doppler parameters and ICP monitoring were performed. Results: Twenty pigs were studied, 10 in group A and 10 in group B. A significant correlation between PI and ICP was obtained, especially shortly after abrupt elevation of ICP. There was no correlation between ICP and FVs, FVd or FVm separately. There was also no significant change in ICP after intravenous infusion of hypertonic saline solution. Conclusions: These results demonstrate the potential of PI as a parameter for the evaluation of patients with suspected ICP elevation.

Invasive ICP monitoring is an important adjunct in the clinical management of ICH, although there are no studies with strong evidence of its benefits 5 . In a trial published in 2012, invasive ICP monitoring was not associated with better patient outcome compared to clinical and tomographic evaluations 6 . Nonetheless, this technique remains the gold standard method for ICP assessment according to the current Brain Trauma Foundation (BTF) guidelines 5 .
With this in mind, techniques based on transcranial Doppler (TCD), a non-invasive method easily accessible at bedside, have been studied. Elevation of ICP leads to changes in the cerebral wave pattern and blood flow velocities obtained by TCD [7][8][9] .
While TCD is a promising method for bedside evaluation, studies that investigated it as a surrogate of ICP have controversial results, especially because of influences of systemic factors on flow velocities and debate concerning pulsatility index (PI) as an indicator of whether ICP or cerebral perfusion pressure. Further studies are needed to confirm this hypothesis [10][11][12] . Thus, in this experimental study, we aimed to correlate cerebral blood flow assessed using Doppler technique with the variations of the ICP in acute phase of intracranial hypertension in an animal model.

METHODS
This experimental study was previously approved by the Ethics Committee for Research Projects of the University of São Paulo Medical School. All applicable institutional and national guidelines for the care and use of animals were followed.

Animals
Hybrid pigs of the Landrace, Duroc, and Pietrain breeds were used. These were brought into the laboratory on the day of the experiment.

Anesthesia protocol
The animals were pre-anesthetized with ketamine (Ketamin-S ® , Cristália) at a dose of 5 mg/kg and midazolam (Dormire ® , Cristália) at a dose of 0.25 mg/kg, both placed in the same syringe and administered intramuscularly. These drugs were selected because they have no significant influence on ICP and cerebral blood flow 13,14 . After 15 minutes, the marginal ear vein was punctured with a 20-or 22-gauge vascular catheter (BD Insyte ® ). Intravenous anesthetic induction with propofol (Provine ® 1% -Cláris) was performed at a dose of 5 mg/kg. The animals were submitted to orotracheal intubation with an endotracheal tube of 6 mm diameter (Portex ® ), and anesthetic maintenance was performed with propofol (Provine ® 1% -Cláris) at a dose of 3 mg/kg/h and analgesia was maintained with fentanyl (Fentanest ® -Cristália) at an initial dose of 5 μg/kg followed by continuous intravenous (IV) infusion of 0.4 μg/kg/min. Neuromuscular blockade was obtained with pancuronium (Pancuron ® , Cristalia) bolus at 0.1 mg/kg IV followed by continuous infusion of this agent at a dose of 0.02 mg/kg/h. After intubation, the animals were submitted to volumecontrolled mechanical ventilation (Dixtal ® 5010 Ventilator). Through an abdominal medial incision, cystostomy was performed under direct vision to control diuresis of the animal. The right femoral artery was punctured and connected to a pressure transducer in all animals for monitoring of invasive mean arterial pressure. Arterial blood gas analysis was performed with samples of 0.3 mL at the beginning of the procedure (in order to establish ventilatory parameters), and after interventions, to evaluate maintenance of the physiological parameters.

Experimental procedure
The ICH animal model developed and previously validated by this research group was used in the present study 15 , inclusive for TCD assessment in swine 16,17 . The model simulates a right frontal intracerebral hemorrhage, performed in a controlled manner. An L-shaped fronto-temporal incision was performed on the head of each animal, at the midline and temporal region just in front of the ear to expose the coronary and sagittal sutures. Then, a bone trepanation 1 cm lateral to the sagittal suture and 1 cm anterior to the coronal suture was made in the right hemicranium, through which the intraparenchymal catheter (Neurovent-P ® , Raumedic ® , Munchberg, Germany) was inserted for invasive ICP monitoring in the frontal lobe white matter. A bone trepanation located 1 cm lateral to the sagittal suture and 1 cm posterior to the coronary suture allowed the introduction of an 8-French pediatric vesical catheter, reaching the subcortical white matter. Then, infusion of 0.9% NaCl solution (PS) was performed for 15 minutes, controlled with infusion pump (Infusomat ® compact, B Braun ® , Melsungen, Germany). A small ipsilateral temporal trepanation was also performed to allow the accomplishment of cerebral Doppler ultrasound with a 5-8 MHz transducer (SonoSite -Micromax, FUJIFILM SonoSite, Washington, DC, United States). This allowed to analyze the cerebral blood flow velocity, establishing the systolic blood flow velocity (FVs), the diastolic velocity (FVd), and from these, the derived parameters were obtained: mean blood flow velocity (FVm=FVs+2xFVd/3) and the pulsatility index (PI) (FVs-FVd/FVm).
The animals were divided into two groups (A and B), in which intracranial hypertension was induced by the inflation of the intraparenchymal balloon with two different volumes, as described below (Table 1). Normal parameters were calibrated in both groups in the first hour.
In group A, the balloon already implanted in the frontal white matter was infused with 4 mL of PS, and soon after, the multiparametric data were collected, which included ICP and the evaluation by TCD. This hematoma is equivalent to an expansion of approximately 80 ml in a human adult brain. This equation is based on the proportion of the brain weight of the animal of 2 months and 18 kg (average of 75 g) relative to normal adult brain weight (1500 g), with a 5% relation. In group B, a 7 mL infusion of PS was performed, equivalent to a volume of approximately 140 mL in a human adult brain 15 .
In both groups, one hour and 30 minutes after onset of balloon inflation, intravenous infusion of hypertonic saline solution (HS; 3% NaCl solution at the dose of 5.3 mL/kg) was performed. After another 30 minutes, we proceeded with balloon deflation, corresponding to the simulation of a surgical procedure.
During the experiment, several parameters were monitored including clinical parameters (pupils), invasive mean arterial pressure (MAP), parenchymal ICP, and TCD measurements (FVs, FVd, FVm, PI) obtained bilaterally from the middle cerebral arteries. These data were collected before and after all interventions on the animals.
At the end of each experiment, the animals were sacrificed through an intravenous dose of propofol (20 mg/kg) and fentanyl (10 mg/kg), followed by 40 mL of 19.1% potassium chloride solution. After the sacrifice, the animals were placed in plastic bags, with labels that clearly identified the origin, content, and the responsible researcher. They were then transported to be incinerated according to our institution routine protocol.

Statistical analysis
Statistical analysis was presented through means and standard deviations, as well as graphs of individual and medium profiles. For each of the measurements, including ICP, adjusted linear mixed regression analysis were applied, considering random effect in the intercept and normal distribution for the random effects 18 . Spearman correlation was calculated for PI and ICP values. The graphical analysis indicated that the random effect of the intercept appeared to differ between groups as well (experiment effect variability in group B was higher than in group A). Therefore, besides considering a random effect of the individual, the effect was considered distinct between groups. The analyses were performed using the R 3.4.0 software (R Core Team, 2017, Vienna, Austria). The results were interpreted using a significance level of 5%.

RESULTS
Twenty two-month-old hybrid pigs with an average weight of 18.46 kg (±1.12) were studied. They were divided in two groups of ten animals: group A (4 males, 6 females) and group B (4 males, 6 females). One pig in group B died before the end of the experiment and was excluded from the analysis. All animals were hemodynamically stable during the experimental procedure, except two animals of group B that presented refractory low blood pressure. Table 2 shows the means and standard deviations observed for ICP measurements collected from the intraparenchymal monitoring and the TCD-based variables FVs, FVd, FVm, and PI. A moderate elevation of ICP was observed in group A and a significant increase was observed in group B after inflation of the balloon (Figure 1). No major ICP variation between the pre-HS and pre-deflation moments were observed in both groups. From the adjusted model, Table 3 was constructed, comparing the groups for each moment and the differences between groups for each evaluated moment (basal, postinflation, pre-HS, post-HS, pre-deflation and post-deflation). There was no statistical difference in ICP between groups before and after the end of the experiment. ICP was higher in group B than group A in the moments just after the insufflation, pre-HS, post-HS and pre-deflation. There was no statistical difference between groups in FVs, FVd, FVm, and PI at any point of the experiment. For transcranial Doppler analysis, data of two animals of group B were excluded due to severe hemodynamic instability ( Figure 2). Hence, subjects without significant changes in systemic hemodynamics during the procedure were accounted for statistical analysis.
There was a moderate correlation between PI and ICP at three moments of the experiment (Spearman correlation coefficients): at baseline (r:0.661), post-inflation (r:0.543), and post-deflation (r:0.578), all with p<0.05. No significant correlation was found between ICP and FVs, FVd, and FVm. The dispersion of the correlations between ICP and PI are presented in Figure 2. It shows that the correlation of ICP with PI   at baseline and shortly after balloon inflation is greater than the correlation between ICP and other PI values over time.
As the elevation in ICP was varied widely among subjects, a precise cut-off was could not be calculated, although Table 2 indicates that for a sudden severe ICH (group B), PI elevation is progressive and less specific for intervention, except for a rapid relief (balloon deflation could simulate decompressive cranioctomy). Figure 3 shows a positive correlation between PI and ICP, especially with substantial elevation in ICP (>30 mmHg). A moderate elevation in ICP tends to respond better to interventions such as hypertonic saline, and PI will have more negative predictive value in these cases.

DISCUSSION
The present study makes important contributions in the development of an animal model with induced and reversible ICH. Correlations between ICP-and TCD-derived parameters were calculated. There was a moderate correlation between PI and ICP at three moments of the experiment. From the pre-HS moment to pre-deflation, there was no significant correlation. The variables FVs, FVd, and FVm were not correlated with ICP at any moment. The data obtained are in agreement with those of some studies and in contrast with others, as discussed below.

Correlation between intracranial pressure invasive monitoring and transcranial Doppler parameters
Invasive ICP monitoring devices have been developed throughout the 20th century and since then have become the gold standard method for this purpose, despite controversial results in some studies 5,6,[19][20][21] . However, due to its potential complications, such as infection, hemorrhage, and misplacement, numerous studies have been conducted in recent years aimed at developing several non-invasive techniques for estimation of ICP 9,22-24 .
TCD is a promising technique with emphasis on PI as a parameter of non-invasive estimation of ICP 9,10,25,26 . TCD is an interesting method because of its availability, portability, and possibility of performing repeated non-invasive tests at bedside.
Bellner et al. investigated the relationship between ICP and TCD PI 9 . They studied 81 patients with various brain lesions (subarachnoid hemorrhage, TBI, and others). It was found a direct relationship between ICP and PI, with a high correlation coefficient (0.938). Similar results were reported in a study of 58 patients with severe TBI treated in the ICU according to the BTF guidelines 27 . Daily TCD was performed for PI comparison with invasive ICP monitoring. There was a strong correlation between PI and ICP, with a 0.779 correlation coefficient in the fifth day.
A retrospective study from Cambridge 10 included 53 patients who underwent an infusion test (consisting of infusing saline solution in the lumbar space, by lumbar puncture, for the study of circulatory disorders). The values of ICP (measured through the lumbar needle) and TCD blood flow velocities were also measured. One of the parameters studied for noninvasive ICP measurement was the PI obtained with TCD. There was statistical significance in the direct correlation between these two parameters (r: 0.45), showing good potential of PI for noninvasive estimation of ICP.
Other studies have also shown positive results regarding the correlation of ICP with PI in patients with brain lesions [28][29][30][31] . These studies are consistent with the results of our experiment in which a strong correlation was found between PI and ICP at three important moments. The correlation coefficient was 0.543 right after balloon inflation, when ICP elevation occurs. At the other moments, when ICP remains high but stable, PI lost its correlation with ICP. We can infer from this that PI may be important in monitoring the worsening of ICP and may be indicated in patients with this suspicion of deterioration.
Bouzart et al. conducted a prospective multicenter study in France in 2016 25 . A total of 356 patients with mild and moderate TBI (Glasgow between 9 and 15) were studied.  Bilateral TCD was performed up to 8 hours post-trauma. The objective was to evaluate TCD as a predictor of secondary neurological deterioration in these patients (which occurred in 6% of study patients) on the seventh day posttrauma. The normal TCD parameter considered was PI less than 1.25 and FVd greater than 25 cm/s. This parameter had sensitivity of 80% and specificity of 79% to predict neurological worsening. The negative predictive value was 98% and the positive predictive value was 18%, suggesting that the normal TCD result is more important in predicting prognosis than the abnormal TCD. Also, this study showed that PI and FVd are related to the prognosis of patients with mild to moderate TBI. This study, despite not evaluating the correlation of the TCD parameters with ICP, reinforces the possible utility of PI in patients with potential risk of neurological worsening.
However, there are many studies in the literature that contradict these positive correlations between ICP and PI. In 2016, another prospective study included 40 TBI patients who were treated in an ICU of a single hospital and who received a parenchymal catheter to monitor ICP 22 . Data were collected from ICP monitoring and TCD. One of the parameters analyzed for non-invasive evaluation of ICP was PI. There was no statistically significant correlation between PI and ICP. The results were consistent with those of Figaji et al., another prospective study that evaluated 34 children with severe TBI, who had ICP monitored 32 . TCD was performed in the middle cerebral artery ipsilateral to the ICP catheter. The aim of this study was to determine the correlation between PI greater than 1 and ICP greater than 20 mmHg, as well as PI less than 1 with ICP less than 20 mmHg. The conclusion of that study was that PI is not a good parameter for noninvasive assessment of ICP in children with TBI. These results are consistent with other published studies 7,22,23,33,34 .

Effect of hypertonic saline solution infusion in the present model
As described in the literature review, hyperosmolar therapy is used to treat cerebral edema and ICH of various etiologies, and mannitol at 20% is the gold standard solution. However, HS, in different concentrations, has been studied for this purpose 35,36 . Hypertonic solutions act through the dehydration of brain tissue and decrease the inflammatory response of the brain to injury, as well as causing positive effects on homeostasis and cardiovascular hemodynamics 37,38 . Although there was no consensus on the best HS concentration for ICH control, in this study the 3% concentration was used because it was equiosmolar to 20% mannitol. In addition, it has shown good efficacy in intraoperative brain relaxation and in controlling ICH of various causes, with good safety and few side effects [39][40][41] .
In the current study, groups A and B maintained stable ICP after HS infusion, without the ICP reduction effect demonstrated in other studies. Perhaps the present animal model of ICH is not adequate to evaluate HS effects. Balloon inflation simulates an acute mass effect with a material that does not respond to changes in blood osmolarity. The benefits of HS are postulated as a result of an osmolar effect, which would not affect the balloon. Other effects of HS as increased cardiac output and inhibition of inflammatory changes are also not applicable in this model. This justifies the results obtained in this experiment, in which there was no change in ICP after infusion of HS. Therefore, the ICH model by balloon inflation simulates a disease process that can only be treated by surgical intervention.  Although the present study makes important contributions in the development of an animal model of induced ICH, it has some limitations. First, Doppler evaluations are highly operator-dependent with a significant learning curve. However, only one accurately trained sonographer performed the Doppler exams to minimize this limitation. Second, FVs, FVd, FVm, and PI parameters are also influenced by blood pressure and blood viscosity. Third, the intracranial solution infusions applied in the study were comparable to extremely elevated intracranial mass volume, which is not the most common situation in clinical practice, although suitable for study purposes.
Additionally, the lesions induced in the study were exclusively performed on the frontal lobe of swines. Theoretically, lesions with the same volume in the posterior fossa may disclose a different behavior on blood flow velocities of middle cerebral arteries. Finally, two animals presented hemodynamic instability, refractory to the stabilization attempts made by the researchers, and were excluded from the PI data analysis. Another limitation of this animal model is the absence of blood contact with brain tissue, with absence of inflammatory reactions caused by a true hematoma.
PI is mostly an indicator of cerebral perfusion pressure, as its formula is based on differences between systolic and diastolic velocities. Previous research used 1.4 42 as the threshold for this index to indicate ICH more accurately. However, in clinical practice, logic leads to individualization, since both intracranial compliance and pressure buffering mechanisms vary from person to person. Rheology, intravascular volume, and the cardiovascular system also play a determining role on cerebral hemodynamics 43 . Thus, the most valuable feature of a non-invasive technique such as TCD may be the opportunity of repeated evaluations and observing particular PI tendency during patient follow-up, associating this with further dynamic variables.
In conclusion, in this experimental study, transcranial Doppler pulsatility index was correlated with ICP monitored by intraparenchymal catheter, especially at the moment of abrupt elevation of ICP. This observation is relevant because similar studies cannot be performed in humans for ethical reasons.