| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J Thorac Cardiovasc Surg 2006;131:395-402
© 2006 The American Association for Thoracic Surgery
Evolving Technology |
a Division of Cardiothoracic Surgery, University of Miami Miller School of Medicine, Miami, Fla
b Pediatric Cardiology, University of Miami Miller School of Medicine, Miami, Fla
c Pediatric Critical Care, University of Miami Miller School of Medicine, Miami, Fla
d Department of Neurological Surgery, University of Miami Miller School of Medicine, Miami, Fla
Read at the Thirty-first Annual Meeting of The Western Thoracic Surgical Association, Victoria, BC, Canada, June 22-25, 2005.
Received for publication May 10, 2005; revisions received June 20, 2005; accepted for publication July 12, 2005. * Address for reprints: Marco Ricci, MD, Division of Cardiothoracic Surgery, University of Miami Miller School of Medicine, Holtz Center 3072 (R-114), 1611 NW 12th Ave, Miami, FL, 33136 (Email: mricci{at}med.miami.edu).
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionDiscussionReferences |
|---|
METHODS: Univentricular physiology was created in newborn piglets by means of an aortopulmonary shunt, echocardiography-guided atrial septostomy, tricuspid valve avulsion, and pulmonary artery occlusion. Intra-aortic balloon inflation was used to increase afterload. Cerebral oxygen saturation monitoring (INVOS 5100; Somanetics Corp, Troy, Mich), physiologic recordings, and stable-isotope microsphere determination of cerebral blood flow were performed at baseline and after conversion to univentricular physiology (30 minutes, 120 minutes, and during afterload augmentation).
RESULTS: Univentricular physiology resulted in lower cerebral oxygen saturation, arterial oxygen content, jugular venous oxygen saturation, and cerebral oxygen delivery. Afterload augmentation increased cerebral oxygen saturation, arterial oxygen content, and jugular venous oxygen saturation, whereas cerebral oxygen delivery was unaffected because of lower cerebral blood flow. Cerebral oxygen saturation predicted jugular venous oxygen saturation, arterial oxygen saturation, and arterial oxygen content. No association was found with cerebral oxygen delivery, which decreased in parallel with cerebral oxygen saturation when the single-ventricle physiology model was established but failed to increase during afterload augmentation.
CONCLUSIONS: This study shows that in univentricular physiology cerebral oxygen saturation correlates well with jugular venous oxygen saturation, arterial oxygen saturation, and arterial oxygen content. However, our findings suggest that in singe-ventricle physiology changes in cerebral oxygen saturation need to be interpreted in the context of changes in arterial oxygenation.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDiscussionReferences |
|---|
It allows the noninvasive detection of changes in the ratio of oxyhemoglobin to deoxyhemoglobin (rSO2 index) in the frontal cortex, thus providing indirect information on the adequacy of cerebral oxygenation.
2
Previous experience with NIRS in clinical practice has focused primarily on the detection of cerebral hypoxia in the context of normoxia and biventricular physiology.
3,4
In this setting rSO2 has been shown to parallel jugular venous O2saturation (jvSO2) consistently.
5
RSO2 monitoring has been found to be useful in a variety of clinical settings, especially cardiac and vascular surgery, as a surrogate indicator of the adequacy of cerebral perfusion.
6,7
Because NIRS is used as a trend monitor, a decrease in rSO2 can reflect a shift in the balance between cerebral oxygen supply and demand and therefore suggests cerebral hypoxia.
Few studies on the use of NIRS in children with congenital heart disease have been reported,
8-10
and the experience with this technology in cyanotic children with single-ventricle physiology (SVP) is limited.
11
In contrast to biventricular physiology, rapid changes in systemic and pulmonary vascular resistance can produce changes in cerebral perfusion in SVP, as well as significant fluctuations in pulmonary blood flow and therefore arterial oxygenation. Previous reports have indicated that NIRS is heavily influenced by markers of arterial oxygenation, such as arterial O2 content (CaO2) and arterial O2 saturation.
12
We hypothesized that changes in arterial oxygenation occurring simultaneously with changes in cerebral blood flow (CBF) could influence the ability of transcranial NIRS to predict changes in cerebral oxygenation. In our study we investigated the relationship between rSO2, as measured with a commercially available NIRS device, and jvSO2 in an animal model of hypoxemia and SVP. We also explored the association between rSO2, cerebral O2 delivery, and other physiologic variables of cerebral oxygenation in an environment in which simultaneous changes in CaO2 and CBF were induced by afterload augmentation.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDiscussionReferences |
|---|
Surgical Preparation
Piglets were anesthetized with intramuscular ketamine (40 mg/kg) and xylaxine (4 mg/kg), intubated through a tracheostomy, and started on volume-control ventilation (tidal volume, 25-30 mL/kg; rate, 25 breaths/min; inspired O2 fraction, 0.25). Anesthesia was maintained with continuous infusion of fentanyl (50 µg · kg–1
· h–1), pancuronium (0.4 µg · kg–1
· h–1), and midazolam (0.2 mg · kg–1
· h–1). A catheter was inserted in the femoral artery for pressure monitoring and blood sampling. A 6F introducer sheath to be used for intra-aortic balloon dilation (afterload augmentation) was inserted in the opposite femoral artery. A 7F introducer sheath was placed in the femoral vein for fluid administration and subsequent insertion of an atrial septostomy catheter. A catheter was inserted in the left internal jugular vein for venous blood sampling. Electrocardiography, rectal temperature, and rSO2 were monitored.
The single-ventricle model resembled that described by others.
13,14
Through a median sternotomy, catheters were placed in the right and left atrium. Heparin was administered (150 U/kg), and a 3.5-mm polytetrafluoroethylene
*
shunt was interposed between the aorta (proximal to the take-off of the innominate artery) and the pulmonary artery. While the shunt was clamped, a 2-mL balloon septostomy catheter (Medtronic Vascular, Danvers, Mass) was advanced from the right femoral vein into the right atrium. Epicardial 2-dimensional echocardiography was used to direct the catheter across the atrial septum and perform a pull-back septostomy. The same catheter was then advanced into the right ventricle. The tricuspid valve was made incompetent by repeatedly withdrawing the inflated balloon across the valve. Finally, the shunt was opened, and the main pulmonary artery was occluded. This allowed the left ventricle to support both the systemic and pulmonary circulations, reproducing a physiology similar to that of pulmonary atresia with intact ventricular septum.
Experimental Protocol
Eight piglets were included in the study group. Animals were ventilated at a constant fraction of inspired oxygen of 25%, adjusting the respiratory rate to maintain a PCO
2 of between 35 and 45 mm Hg. Rectal temperature was kept at 35.5°C to 36.5°C. On the basis of experience from previous pilot experiments, normal saline (4 mL · kg–1
· h–1), dopamine (5-10 µg · kg–1
· min–1), and epinephrine (0.05-0.1 µg · kg–1
· min–1) were administered in all animals throughout the entire experiment to maintain cardiovascular stability and avoid hemodynamic perturbations. Calcium gluconate and sodium bicarbonate were given as needed. Fresh whole blood was obtained from an adult pig and infused to maintain a hemoglobin concentration as close to baseline as possible.
Data collection was carried out at baseline and after conversion to SVP (30 minutes, 120 minutes, and during afterload augmentation). Measurements during afterload augmentation were obtained 15 minutes after inflating a balloon in the distal descending thoracic aorta. Data collected included hemodynamic parameters, blood sampling (arterial, central venous, and cerebral venous, hemoglobin, and lactate), determination of total cardiac output with an electromagnetic flowmeter, determination of CBF with stable-isotope microsphere injections, and transcranial measurement of rSO2 by using NIRS. At completion, piglets were killed with KCl and fentanyl. An autopsy was performed to confirm the correct positioning of all indwelling catheters and the adequacy of the atrial septostomy and that of the aortopulmonary shunt anastomoses. The brain was removed and weighed for blood flow determinations.
Determination of CBF
Stable-isotope microspheres (15 ± 5 µm; BioPhysics Assay Laboratory, Inc, Worcester, Mass) were used as described by others.
15
Microspheres labeled with nonradioactive isotopes were administered over 3 seconds through the left atrial catheter, and reference samples were collected over 90 seconds from the femoral artery (2 mL/min). A total of 1 x 106 microspheres was injected in the biventricular state, whereas 2.5 x 106 were injected in SVP. Experience with microspheres in a univentricular model is lacking in the literature. On the basis of our previous experiments, we demonstrated that 2.5 x 106 microspheres in the univentricular state compensated for run-off through the pulmonary circulation and ensured adequate microsphere concentration in the brain. At the end of the experiment, tissue samples were obtained from different brain regions (cortex, white matter, thalamus, hippocampus, brainstem, and cerebellum), weighed fresh, and dried in a warming oven at 60°C for 48 hours. Dried samples were sent for analysis and processed by using neutron activation.
15
This technique entails exposure of the isotope to a neutron beam, which results in an activated radioactive nucleus. Activated labels are then allowed to decay for 48 hours, during which gamma rays are emitted and measured. The signal is proportional to the concentration of the isotope in the sample. Results of the assay for each label are reported in disintegrations per minute (dpm). CBF is expressed in milliliters per minute per 100 grams and calculated by normalizing the concentration of microspheres in the tissue sample (dpm per gram) to the concentration in the reference sample (dpm per minute per milliliter). Total CBF was calculated by averaging all measurements obtained from both hemispheres.
Determination of rSO2
A commercially available device (INVOS 5100 Cerebral Oxymeter; Somanetics Corp, Troy, Mich) was used to measure the tissue oxygen saturation index (rSO2 index) in the frontal cerebral cortex.
5,8
This technology entails the use of near-infrared light (730-nm and 810-nm wavelength) to measure the absorption spectra of oxyhemoglobin and deoxyhemoglobin. After shaving the scalp, a 30-mm patch containing the NIRS-emitting source and sensor was placed on the forehead and connected to the infrared spectrophotometer unit. The NIRS probe measures the rSO2 index in a banana-shaped volume of tissue of 1.5 cm3 in depth, subtracting superficial light-scattering signals that originate from extracranial tissues. The device monitor displays data continuously, although for the statistical analysis, we used NIRS data collected at steady state at selected data points simultaneously with arterial and jugular venous blood gases.
Physiologic Measurements and Calculations
Total cardiac output (excluding coronary blood flow) was determined by using an electromagnetic flowmeter (Transonic Systems, Inc, Ithaca, NY) placed on the ascending aorta. After conversion to SVP, pulmonary blood flow was determined by placing the flowmeter on the pulmonary artery distal to the aortopulmonary shunt, as described by others.
13,14
Total left ventricular outflow indexed by body weight was calculated as left ventricular output (in milliliters) per minute divided by body weight (in kilograms). The pulmonary/systemic blood flow ratio was calculated as pulmonary blood flow (in milliliters per minute) divided by systemic blood flow (excluding coronary flow, in milliliters per minute). CaO2 was calculated as follows: . Cerebral O2 delivery (in milliliter of O2 per milliliter) was calculated as the product of CBF (in milliliters per minute per 100 g) and CaO2 (in milliliters of O2 per 100 mL). The arteriovenous O2 saturation difference was calculated as the difference between arterial O2 saturation and jvSO2.
Statistical Analysis
All physiologic data are expressed as means ± SEM. Normal distribution of variables was assessed with the Kolmogorov-Smirnov normality test. Relations between rSO2 and other variables were determined by means of least-squares regression analysis and calculation of Pearson correlation coefficients. One-way repeated-measures analysis of variance was used to compare physiologic data obtained at different data points during the experiment (baseline, 30 minutes, 120 minutes, and during afterload augmentation).
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDiscussionReferences |
|---|
|
|
illustrate the statistical relationship between rSO2 and other physiologic variables. Linear relations were observed between rSO2 and arterial O2 saturation (Figure 1), jvSO2 (Figure 2), and CaO2 (Figure 3). Conversely, rSO2 failed to predict cerebral O2 delivery (Figure 4) and CBF (Figure 5).
|
|
|
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDiscussionReferences |
|---|
Detection and prevention of cerebral hypoxia might be critical to avoid adverse neurologic outcomes in children with congenital heart disease.
8
Cyanotic children with SVP might be especially vulnerable to cerebral hypoxia and brain injury because they are hypoxemic and have a limited cardiac output reserve.
16
NIRS technology has recently emerged as an effective and noninvasive tool to monitor brain oxygenation perioperatively. Clinically, NIRS-based brain monitoring with algorithms for interventions has been proposed as a tool to reduce the incidence of postoperative neurologic complications both in adult cardiac and congenital cardiac surgery.
Oxygen transport to the brain is a complex phenomenon influenced by several physiologic variables. These might have an effect on rSO2 and on its ability to detect cerebral hypoxia.
17
Oxygen delivery to the brain is determined by CBF and CaO2, which in turn is influenced by hemoglobin concentration and arterial O2 saturation. NIRS has been shown to monitor oxygen saturation in a selected region of cerebral cortex.
18
Previous work has demonstrated that rSO2 parallels closely cerebral venous saturation because of the overrepresentation of the venous capillary component over the arterial component in the cerebral tissues.
17,18
Kurth and colleagues,
18
as well as others,
12,17
found a significant correlation between cerebral saturation measured by using NIRS technology and cerebral venous saturation. In the setting of biventricular physiology and steady pulmonary function, arterial O2 saturation and CaO2 are unlikely to change, at least over short periods of time. If cerebral metabolism and oxygen consumption remain constant during the same time interval, changes in cerebral venous O2 saturation and therefore rSO2 can be assumed to reflect changes in CBF and ultimately cerebral O2 delivery. On the basis of this, rSO2 monitoring has been found to be useful in clinical practice because it can provide information on the adequacy of cerebral perfusion.
3-5
In SVP, however, changes in arterial O2 saturation can occur rapidly as a result of the varying balance between systemic and pulmonary vascular resistance.
19
These could affect rSO2 and interfere with the ability of NIRS to reflect changes in the relationship between cerebral O2 supply and demand. As Table 1 shows, in our experimental model the rSO2 index decreased in parallel with cerebral O2 delivery and arterial O2 saturation as the SVP model was established. However, this correlation was lost when afterload augmentation was introduced because rSO2 increased significantly, whereas cerebral O2 delivery did not. In our experiment afterload augmentation caused a significant increase in arterial pressure and pulmonary blood flow through the aortopulmonary shunt, which in turn resulted in higher arterial O2 saturation and lower PCO
2. These physiologic changes caused a modest decrease in CBF. The net result of the higher CaO2 and lower CBF was that cerebral O2 delivery remained unaffected, whereas rSO2 increased significantly, tracking the increasing arterial O2 saturation.
Continuous monitoring of rSO2 allows rapid recognition of cerebral hypoxia
20
and immediate feedback on the efficacy of therapeutic maneuvers aimed at improving cerebral oxygenation.
3
The use of NIRS for brain monitoring has been especially studied in the setting of cardiac operations. In 250 children undergoing operations for congenital heart disease, Austin and coworkers
8,9
found that using NIRS brain monitoring to guide interventions to improve cerebral oxygenation resulted in a decrease in the incidence of neurologic sequelae and shorter hospital stay. Similarly, NIRS has been proposed as a useful tool to monitor brain oxygenation both preoperatively
10
and postoperativly,
11
although the experience in these settings is more limited. The implications of our findings as they relate to rSO2 monitoring in children with SVP could be significant in that they unveil a physiologic situation in which the use of rSO2 as a surrogate indicator of the adequacy of cerebral perfusion might have limitations. The response to afterload augmentation noted in our piglets suggests that NIRS data need to be interpreted with caution if simultaneous changes in arterial O2 saturation are occurring. These should be taken into account when making assumptions on the efficacy of therapeutic maneuvers in improving brain oxygenation.
Our findings must be interpreted in light of the limitations inherent with the experimental model. Despite the fact that this model has been validated previously, it is unknown whether our experimental conditions reproduce cardiovascular changes occurring in children with SVP. Differences between animal species and human subjects could also influence rSO2 monitoring with NIRS. Additional limitations relate to the method used to increase afterload (ie, balloon inflation in the descending thoracic aorta). We used aortic occlusion because this method has been validated by others
21
and is highly reproducible. In addition, because our piglets were already receiving inotropic support during the entire experiment, further increasing afterload through additional pharmacologic interventions would have been impractical. Nonetheless, we acknowledge that aortic occlusion might have produced physiologic changes in the SVP model that are not directly equivalent to those observed in clinical practice when afterload is manipulated pharmacologically. Also, the fact that afterload increases were produced 2 hours after creation of the model might have been associated with some deterioration of the experimental preparation.
In our study we did not measure intracranial pressure, and therefore we cannot draw conclusions on the effect of this variable on rSO2 changes. It is also possible that afterload augmentation induced changes in the cerebral circulation that affected the balance between arterial and venous components as detected by the NIRS signal, therefore influencing rSO2 values. Additional weaknesses relate to the lower hemoglobin concentration found in piglets, unlike the relative polycythemia seen clinically, and to the use of jugular venous saturation as opposed to sagittal sinus saturation to estimate cerebral venous oxygenation.
Despite the many limitations, this study could provide additional information on the use of NIRS in children with univentricular physiology. NIRS technology is useful because it allows the rapid and noninvasive detection of cerebral hypoxia and the assessment of responses to therapeutic maneuvers aimed at improving cerebral oxygenation. Our study suggests that in the setting of SVP, rSO2 trends need to be interpreted in the context of changes in arterial O2 saturation. Our findings do not diminish the value of transcranial NIRS in assessing cerebral oxygenation. However, they provide a useful adjunct to the interpretation of rSO2 changes when simultaneous changes in arterial oxygenation are occurring.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDiscussionReferences |
|---|
Dr Ricci. Thank you for your comments. I share your view on the difficulty of the model. This model is not only difficult but also very humbling and we certainly lost a lot of animals in trying to refine it and get to this point.
Of course NIRS has been proposed and has been used with increasing frequency at many centers including ours, because it might provide some information on the balance between supply and demand and, therefore, on the adequacy of cerebral oxygenation. However, it has limitations. Our hypothesis was that, as NIRS is essentially a very reliable indicator of cerebral venous oxygenation, changes in arterial oxygenation could influence the NIRS signal and could affect the ability of this technology in providing information on cerebral oxygenation. Obviously after we got into our conclusions, our impression is that in SVP, since there is a possibility that the CaO2 might change in contrast to normal biventricular physiology, simultaneous changes in arterial saturation need to be taken into account when interpreting NIRS changes.
As far as the extent to which these findings would apply and change our management in clinical situations, I cannot answer that question. I can tell you that we are using this technology now in our pediatric intensive care unit for all our pediatric patients, including patients with SVP. We began using it relatively recently, so I do not have clinical data to share with you regarding this. However, knowing what we know now, let's say we see that the cerebral saturation by NIRS had declined by 20% in a child with SVP. Based on this and other clinical indicators, we would institute therapeutic maneuvers to try to correct that, with the assumption that cardiac output may be low. Then, we know that if we see a rise in cerebral oxygenation as shown by NIRS in response to the intervention, and that, for example, the NIRS increase corresponds to a simultaneous rise in arterial oxygenation, then that NIRS rise might not necessarily indicate better cardiac output in response to the intervention. It could simply reflect improved arterial oxygenation. I do not know whether the message is clear, but I think these findings insert an additional element in the equation when making clinical decisions based on NIRS, which I think is important.
Dr Doty. So the question is, does everybody who does this kind of work need to run out and buy one of these things tomorrow, or are you still content with getting along with what you had before, which is nothing?
Dr Ricci. I don't think I can answer that question based on these data.
Dr John Meyer (Boston, Mass). It seems to me that the longstanding problem with using this technology has been trying to figure out what is happening to the blood that is in the pathway of your NIRS signal versus what is happening in the parenchyma. What your information would indicate to me, or the way I would interpret it, is that the NIRS technology is picking up what is happening in the blood, not what is happening in the cerebral parenchyma. The problem of trying to separate out those two contributors to your NIRS data has been a difficult one. I think actually this is pretty good information to suggest that the NIRS signal may not be reflecting what is going on in the cerebral parenchyma, the neuronal cells, and is probably more reacting to or a function of what's happening in the blood that happens to be in the pathway between the signal. I think from that standpoint this is actually an extremely useful study and I congratulate you on doing it.
Dr Ricci. We used 0.25.
Unidentified Speaker. A number of years ago, in the days when we were letting babies sit around waiting for a heart transplant, a colleague of mine in Salt Lake City looked at carotid blood flow but at lower oxygen saturation and it just drove home the physiologic point. He found that the flow, and actually what he calculated was a crude measure of oxygen delivery to the brain, was better at a lower FIO 2 than it was higher. I guess it just drives home the point that oxygen delivery is ultimately determined by hemoglobin, oxygen saturation, and flow. Once again we learned the foundation of physiology here. Thank you.
Dr Ricci. Thank you.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
* Gore-Tex shunt, registered trademark of W.L. Gore & Associates, Inc, Newark, Del. 
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
