| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
From the *Klinik für Anästhesiologie und Intensivmedizin, Universität Duisburg-Essen and Universitätsklinikum Essen; and Klinik für Transplantations-und Allgemeinchirurgie, Universität Duisburg-Essen and Universitätsklinikum Essen, Essen, Deutschland, Germany.
Address correspondence and reprint requests to Dr. Simon T. Schäfer, Klinik für Anästhesiologie und Intensivmedizin, Universitätsklinikum Essen, Hufelandstr. 55, D-45122 Essen, Deutschland. Address e-mail to simon.schaefer{at}uni-due.de.
Abstract
BACKGROUND: Venous air embolism (VAE) is a potentially fatal complication during surgical procedures with patients in the sitting position. Since methods for detection of persistent low-volume VAE and targeted air aspiration are limited, we tested the hypotheses that transvenous intracardiac echocardiography (ICE) 1) improves detection of small air emboli in comparison to transesophageal echocardiography (TEE) and precordial Doppler monitoring (PCD) techniques, and that 2) image-guided multiorifice central venous catheter manipulation improves air recovery in moderate and large VAE, when compared with aspiration with the multiorifice central venous catheter in a static position.
METHODS AND RESULTS: Adult swine (73 ± 4.6 kg, n = 7) were premedicated, anesthetized with propofol and fentanyl, endotracheally intubated, mechanically ventilated, and placed in a 45° head-up position. First, nine different small volumes of air emboli (0.05–1 mL) were randomly injected via an ear vein, and VAE detection methods were applied in random order. For 378 small volume air injections, ICE had a much higher sensitivity (82.5%, P < 0.0001) on the analysis of VAE detection than TEE (52.8%) or PCD (46.8%), with no difference (P = 0.571) between TEE and PCD. An injected air volume as small as 0.15 mL was detected by ICE in 90% of injections performed, whereas PCD and TEE detected only half of the boluses of 0.25–0.30 mL of air, and required boluses of 0.4–1.0 mL to achieve 100% detection. Air recovery was assessed in a second series of moderate VAE (2, 5, 10 mL); image-guided aspiration-catheter manipulation recovered significantly more (34.1% vs 17.2%, P < 0.0001) intracardiac air than without catheter manipulation. In a third series of injections of large air volumes (25, 50, and 100 mL), air recovery was not significantly different with ultrasound-guided aspiration (41.3% vs 31.8%, P = 0.11).
CONCLUSION: Small air emboli are detected by ICE with much greater sensitivity compared with both PCD and TEE techniques. Furthermore, recovery of embolized air is enhanced by image-guided manipulation of a multiorifice central venous catheter. Clinical studies are required to assess this technique during surgery with patients in the sitting position.
Venous air embolism (VAE) is a frequent and potentially fatal complication during surgery with patients in the sitting position, and may occur whenever venous pressure at the site of surgery is near or below atmospheric pressure.1–3 Air entrainment can present as either acute massive air embolism, evoking immediate hemodynamic compromise, or as persistent low volume VAE, which may be prodromal to massive VAE. Furthermore, small amounts of air, which might be below the detection threshold of transesophageal echocardiography (TEE) or precordial Doppler (PCD) techniques, may embolize continuously over a longer time into the pulmonary circulation, evoking cardiopulmonary disturbances due to their cumulative effect.4–6
The monitoring techniques available for the detection of VAE include TEE, PCD, and changes in pulmonary artery pressure (PAP), end-tidal carbon dioxide tension (ETco2), or central venous pressure. At present, TEE is considered the most sensitive method for detecting VAE with a threshold of 0.02–0.19 mL air per kilogram body weight,7–9 followed by PCD monitoring with a detection sensitivity reported as ranging from 0.05 to 0.24 mL/kg.8,10 Detection sensitivity is less, and ranges from 0.25 to 0.76 mL/kg, for changes in PAP, ETco2 tension, and central venous pressure.3,8,11 In fact, changes in PAP, ETco2, or central venous pressure only represent the sequelae of pulmonary vascular obstruction by embolized air. Thus, traditional monitoring methods require relatively large volumes of embolized air for detection of VAE, and persistent small volume VAE might not be detected until major hemodynamic compromise occurs. Therefore, strategies to improve early detection of VAE, in particular, in the small volume range, must focus on detection of intracavitary air.
Furthermore, once VAE occurs, immediate action is required to minimize hemodynamic disturbances. Besides the earliest possible recognition of VAE and the prevention of further intravascular air entry, the only definitive therapy is the aspiration of embolized air via a multiorifice central venous catheter (CVC). Air recovery varies widely from 24.2% to 80%, depending on injected air volume and injection site.12–15
Intracardiac echocardiography (ICE) provides a unique high-quality image of the right heart cavities and is routinely used in cardiology for guidance of ablation catheters or patent foramen ovale closure procedures.16–19
Thus, we tested the hypotheses that ICE improves 1) detection of small air emboli, and 2) recovery of intracardiac air after larger volume VAE. Accordingly, we compared the sensitivity for small volume (0.05–1 mL) VAE detection of ICE to both TEE and PCD monitoring techniques. Furthermore, we analyzed whether ICE image-guided aspiration of embolized air in moderate (2–10 mL) and large (25–100 mL) VAE via a multiorifice CVC is superior to aspiration with the CVC in a static position.
METHODS
The experimental protocol was approved by the local animal care committee (TSG-No.: G851/05), and animals were treated in accordance with the guidelines of the American Physiological Society and the Guide for the Care and Use of Laboratory Animals (National Institute of Health publication 85-23, revised 1996).
Animal Preparation
Seven male pigs (weight: 73 ± 4.6 kg sd) were studied. After an overnight fast, the animals were preanesthetized with ketamine (27 mg/kg), azaperone (2.2 mg/kg), and atropine (2 mg IM). After placement of an 18-guage peripheral line into a left ear vein and injection of propofol (2 mg/kg), the animals were endotracheally intubated and mechanically ventilated (rate: 12/min, tidal volume: 8 mL/kg, inspiratory oxygen fraction: 0.5). Anesthesia was maintained by propofol (7 mg · kg–1 · h–1) and fentanyl (5 µg · kg–1 · h–1) IV, and 4 mL · kg–1 · h–1 lactated Ringer’s solution was infused continuously via the peripheral line used for air injections.
For continuous arterial blood pressure monitoring, an 18-gauge catheter (Vygon GmbH, Aachen, Germany) was inserted into the right carotid artery. An 11F sheath (Arrow International Inc., PA) was inserted into the right jugular vein for placement of a 10F ICE catheter (AcuNav, Siemens, Erlangen, Germany). Additionally, via 8.5F sheaths (Arrow International Inc., Reading, PA), using the right jugular vein, a 7.5F pulmonary artery catheter (CritiCath. SP5507 TD Catheter, Becton Dickinson Inc., Sandy, UT) and a 7.5F single-lumen multiorifice CVC (Arrow International Inc., Reading, PA) were inserted. The multiorifice CVC has a length of 30 cm, a maximum flow rate of 5300 mL/h, and three openings, located at the tip, and 1.4 and 2.7 cm away from the tip, respectively.
The CVC was advanced via the jugular vein and position of its tip at the atriocaval junction was assumed when the p-wave amplitude of the intracardiac electrocardiogram (ECG) started to increase.20,21 An intraatrial ECG was obtained by connecting lead II to the external end of the saline filled single-lumen CVC with standard limb leads being connected to the animal, as described previously.22 Afterwards, the upper right atrium was visualized by ICE and the CVC was positioned with its tip within the upper 20 mm of the right atrium, as this position is reported being optimal for aspiration of venous gas emboli.13,23
Measurements
A commercially available ultrasound machine (Sequoia C256, Siemens, Erlangen, Germany) was used for all echocardiographic studies.
Transvenous ICE
ICE was performed using a 10F catheter with a 5–10 MHz multiple frequency transducer (AcuNav, Siemens, Erlangen, Germany). To ensure standardized simultaneous imaging of the right atrium and ventricle, the catheter was placed into the upper right atrium and rotated until the right atrium and ventricle were clearly visible, as described previously (Fig. 2A).24
|
TEE
TEE was performed using a multiplane 3.5–7 MHz probe (SK V5 M, Siemens, Erlangen, Germany). The probe was advanced into the esophagus and rotated until the right heart was clearly visible, and angulation was used to enhance right ventricular imaging at a midesophageal level.25
PCD
A PCD probe (915-AL, Parks Medical Electronics, OR) was positioned to the left of the sternal prominence,26 and a characteristic Doppler sound change after a 2-mL injection of air was required to ascertain an adequate positioning, otherwise the PCD probe was placed more laterally. Thus, in all experiments, the Doppler probe position was suitable to detect intracardiac air, if present.
Data Acquisition, Storage, and Analysis
Both assessors completed practical and theoretical training regarding ICE imaging and catheter manipulation techniques at our institution prior to the study. The operator who manipulated the ICE is an echocardiographer with experience in TEE, ICE and transthoracic echocardiography, and he additionally passed a practical course on ICE in swine before the experiments. To imitate the clinical setting, real-time analysis was performed.
Echocardiographic studies were stored on video tape (Sony medical systems SVO-9500MDP, Sony Electronics Inc., San Jose, CA). Additionally, digital loops were stored on a hard drive for possible off-line analysis using the Image Arena 2.8.1 Software package (TomTec, Unterschleissheim, Germany).
For synchronization of ultrasound images and hemodynamic variables, the device clocks were mechanically adjusted and a 0.1-Hz synchronization signal was displayed on the ultrasound screen using a wave generator27 (Arbitrary Wave generator 33120A, Hewlett Packard, Loveland, CO) and simultaneously recorded on a thermoarray recorder (DASH 8x, AstroMed. Inc., Warwick, RI), but these data were not used for the data analysis reported in this article.
Cardiovascular Variables
Arterial blood pressure, central venous pressures, PAP, and ECG were obtained using transducers (SC 9000, Siemens, Erlangen, Germany). Pressures were referenced to atmosphere and recorded at a midchest level, being defined as the center point of a line drawn from the sternal prominence to the examination table. ETco2 was measured using a capnograph (Capnomac Ultima, Datex Ohmeda, Helsinki, Finland). All signals were continuously stored on Tape (RD-135T, DAT DATA Recorder, TEAC Deutschland GmbH, Wiesbaden, Germany), and recorded on the thermoarray recorder.
Experimental Protocol
After instrumentation, animals were placed in a 45° head-up position, as described.14 First, we determined the sensitivity of ICE, PCD, and TEE techniques for the detection of small air bubbles with detection techniques applied in random order (Fig. 1). Nine increments of air (0.05–1 mL) were injected via a left ear vein in a randomized order, and each air volume was injected twice for each method of monitoring and animal, resulting in 54 low-volume air injections per animal. Injections were separated from another by at least 120 s, i.e., when Doppler sounds had returned to normal and no more intracardiac air could be visualized by TEE or ICE, respectively. To minimize investigator bias, injections were performed at any time within a 60-s window, with the assessors being unaware of the timing of injection. Two independent assessors then categorized detection of VAE as positive, equivocal, or negative. A positive detection was defined as a clearly visible intracardiac air bubble or a characteristic Doppler signal change, respectively. An equivocal detection was assumed when the assessors thought they recognized a suddenly appearing, echodense, intracardiac structure, which appeared atypical for a venous air bubble or Doppler sound change, respectively, but were uncertain whether this was an artifact. A negative detection was defined as the absence of any recognizable intracardiac air or a Doppler sound change. If assessors could not agree on the presence of an air bubble, the result was classified as equivocal.
|
ETco2, mean PAP, and central venous pressure were documented before and 1 min after every air injection.
Second, after completion of the sensitivity studies, we analyzed intravascular air recovery after injection of moderate (2–10 mL) and large (25–100 mL) air volumes using two aspiration techniques via a multiorifice CVC previously placed into the upper right atrium. The catheter was either advanced and withdrawn under ICE image guidance during aspiration, i.e., when intraatrial air was visible, or kept in a static position while aspiration was performed.
For these experiments, 2-, 5-, and 10-mL injections of air were performed in a random order, and every injection and aspiration was performed twice for each aspiration method. Subsequently, large air emboli of 25, 50, and 100 mL were injected in an ascending order as the final part of the protocol, and these injections were not repeated. Every air injection was conducted over a 5-s period, and once VAE was confirmed, aspiration via the multiorifice CVC was started using a syringe. Aspiration was aborted when no more intraatrial air could be aspirated. The amount of air recovered was visually determined by two independent investigators after foam in the syringe had settled and recorded. When no more intracardiac air could be visualized by ICE, the subsequent injection was performed. Whenever hemodynamic instability occurred, no further air injections were performed and the study was terminated.
Statistical Analysis
Data are presented as means ± sd, unless indicated otherwise. Sensitivity of VAE detection is expressed as the percentage of positive detections related to the total number of small volume air injections, and by plotting detection frequency against air volumes injected.
Comparisons between ICE, TEE, and PCD were performed using the nonparametric Sign Test for related samples, as this allows testing for a trend in a series of ordinal measurements as well as testing for a correlation.
Effects on mean PAP, central venous pressure, and ETco2 of VAE were analyzed using the Wilcoxon-test. Linear regression analyses were performed for each experiment, and averaged linear equations were calculated for cumulative air dosages administered with aspirated air subtracted.
Additionally, we assessed the individual dose of air leading to circulatory breakdown. Therefore, the individual lethal dose was calculated by subtracting the amount of air aspirated from the sum of air injected.
Recovery of embolized air is expressed as the percentage of the amount of air injected. The Wilcoxon-test was used to assess differences between results of aspiration techniques. Statistical analysis was performed using the SPSS 13.0 Software package (SPSS Inc., Chicago, IL) and the Microsoft Excel software package (Microsoft Office XP, Microsoft, WA). An 
-error P < 0.05 was used to indicate statistical significance.
RESULTS
Sensitivity of Detection Methods in Small Volume VAE
Figure 2A shows the standardized view obtained by ICE used for all experiments, and Figure 2B is a representative example of a venous air embolus detected by ICE after a 0.5 mL air injection.
ICE was significantly (P < 0.0001) more sensitive than TEE or PCD in detecting small amounts of embolized air after 0.05–1 mL air injections, with an overall (n = 378 injections) incidence of positive detection of 82.5% compared with 52.8% and 46.8%, respectively (Fig. 3).
|
Furthermore, with air volume injected increasing from 0.05 to 0.2 mL, there was an early steep increase in detection sensitivity from 28.6% to 100% with ICE, whereas sensitivities for TEE or PCD increased less with increasing air dosages, and the air volume detection relationships for TEE and PCD were shifted to the right compared with ICE (Fig. 4, Table 1). Thus, the 90% detection dose was 0.15 mL for ICE, whereas 0.4 mL air emboli was required for PCD and 1.0 mL for TEE, respectively.
|
|
The detection rate for VAE was significantly lower (P < 0.0001) with TEE and PCD but not statistically different between TEE and PCD (P = 0.571).
IV air injections in dosages of 0.05–1.0 mL evoked no significant changes in PAP, ETco2, and central venous pressure.
Recovery of Embolized Air in Moderate and Large VAE
ICE image-guided CVC manipulation enhanced air recovery in VAE in comparison to aspiration with the CVC in a static position. Overall (n = 112 injections), recovered air volume averaged 34.9% with ICE guidance and 20.7% with standard aspiration (P < 0.0001) over the complete dose range (2–100 mL) assessed. All seven animals survived the 25-mL experiments, whereas six animals completed the 50-mL injections, and only one animal the 100-mL injections, respectively.
Average air recovery increased with higher air volumes injected. When comparing air recovery after the injection of 2 mL and 100 mL of air, the average recovery increased from 30% to 56% with ICE guided aspiration, and from 11.0% to 48% with conventional aspiration (Fig. 5).
|
In particular, image-guided aspiration yielded a significantly higher air recovery after moderate dosages (2, 5, or 10 mL) of air injected (34.1% vs 17.2%; P < 0.0001) than the conventional aspiration technique (Fig. 5). Experiments with larger doses of injected air (25–100 mL) showed an insignificant difference in air recovery with ultrasound-guided aspiration (41.3% vs 31.8%. P = 0.11).
Dose-Dependent Effects of VAE on Cardiopulmonary Variables
Mean PAP at baseline averaged 13.8 mm Hg and increased linearly with the cumulative dose of injected air (Fig. 6). ETco2 tension at baseline averaged 31.9 mm Hg and declined in a linear fashion with increasing amounts of air injected. The average volume of venous air embolism evoking circulatory breakdown was 2.31 mL/kg (±0.58 mL).
|
DISCUSSION
This study is the first to assess the feasibility and sensitivity of ICE for the detection of VAE. Our data show that ICE is much more sensitive in detecting small air emboli than both TEE and PCD. Furthermore, ultrasound-guided manipulation of the multiorifice CVC significantly enhances air recovery with moderate dosages (2–10 mL) of air injected IV.
During surgery, with patients in the sitting position, air embolism is frequent1–3,28–30 and clinical experience shows that VAE either occurs suddenly, causing immediate hemodynamic compromise, or appears as more prolonged small volume air entrainment, initially evoking little cardiopulmonary compromise.6,31 In any case, alterations in mean PAP or ETco2 tensions caused by either massive or prolonged small volume air entrainment indicate that substantial VAE has already occurred. Accordingly, these latter methods for monitoring are insensitive, with a detection threshold as high as 0.25–0.76 mL/kg of air,3,8,11 and only represent the aftermath of passed emboli. Furthermore, Drummond et al.32 showed that responses are delayed with the maximum changes in PAP and ETco2 occurring after 0.92 ± 0.7 min and 1.85 ± 0.7 min after an 1.5 mL/kg air injection, respectively. Monitoring these variables is potentially helpful for guiding cardiovascular therapy subsequent to VAE, but not for air removal.
Thus, an early warning system is desirable for VAE to detect small air emboli, allowing the clinician to take immediate action before full-blown VAE with pulmonary hypertension and right heart failure develops.
Currently, TEE and PCD are used for this purpose and reported detection sensitivities of both methods vary widely and overlap (0.02–0.19 mL/kg for TEE and 0.05–0.24 mL/kg for PCD).5,8–10 Our data confirm similar detection sensitivities for TEE and PCD, with a detection threshold being lower than reported in the literature (50% detection rate 0.004 mL/kg for TEE and PCD) but higher than for ICE (0.002 mL/kg). The higher detection sensitivities might have been due to technical improvement of monitoring devices. Nevertheless, detection sensitivity might be over-estimated in comparison to clinical practice, as the assessors were alert and expected upcoming VAE. However, using the same protocol and setting for all detection techniques, a bias for a specific technique of detection used for VAE can be excluded.
However, the feasibility of TEE in patients undergoing surgery in the sitting position is limited, since accessibility is reduced because of anteflexion and torque of the head, and surgery may need to be halted during TEE probe manipulation to avoid vibration-related complications. Furthermore, major complications, such as compression-induced tissue damage, tongue necrosis, esophageal varical bleeding,33,34 and endotracheal tube dislocation,35 have been reported. Doppler probe sensitivity, in turn, depends highly on successful probe positioning with recommended positions varying from left to right parasternal11,15 and a detection sensitivity reported to range from 29% to 88% with a 1-mL injection of CO2.11,15,26
ICE is a technique16 routinely used in interventional cardiology and provides a superior image quality of cardiac structures.19,36–38 Although no major procedural-related complications, no evidence of thrombus formation on the catheter, or damage to cardiac structures have been reported, further studies are needed to analyze feasibility and safety of ICE in the operating room.16,39 Manipulating an ICE probe is easier due to smaller catheter diameter and the possibility of advancing the catheter either via the internal jugular vein or the femoral vein. Presumably, transfemoral access is more suitable in patients undergoing neurosurgical procedures in the sitting position, but further studies are needed.
Our data show that ICE is exquisitely sensitive for detection of VAE, detecting air emboli as small as 0.15 mL with a sensitivity of 92.9%. In contrast, conventional TEE and PCD techniques proved far less sensitive for detection of small emboli. This suggests that ICE could be a very useful early warning tool in small volume VAE. This is important, since small emboli may be prodromal to massive VAE.31 Furthermore, persistent infusion of even very small air bubbles (0.3–1 mm diameter) into the pulmonary circulation provokes endothelial damage resulting in increased permeability,4,40–42 and producing an increased risk for pulmonary dysfunction and edema,43,44 due to interstitial neutrophil infiltration, complement activation, and lung injury.45–48
Treatment of VAE includes communication with the surgeon to minimize further air entrainment and the anesthesiologist’s effort to aspirate embolized air via a multiorifice CVC previously positioned by intracardiac electrocardiography. Air recovery is reported as 24.2% to 80%,12–15 and essentially depends on air entrainment per time, detection method used, and the type of aspiration catheter used.12–15 To enhance air recovery, we used ICE, which we found to be the most sensitive method for intracardiac air detection, to visually guide catheter manipulation and air aspiration after VAE. Image-guided advancement and withdrawal of a multiorifice CVC within the right atrium significantly enhanced air recovery, in particular, with moderate air volumes (2–10 mL) injected. Presumably, reducing the amount of air, which finally reaches the pulmonary circulation, enhances patient safety, and therefore, improving air recovery even in the 2–10 mL range might be of clinical importance. Thus, from a clinical perspective, the potential value of ICE, besides early detection of VAE by its exquisite sensitivity, likely rests in the enhancement of intracardiac air removal to migitate potential circulatory depression. The observation that ICE image-guided aspiration after the injection of 25–100 mL of air only showed a nonsignificant trend towards higher air recovery might have been due to very high initial intraatrial air accumulation, resulting in enhanced aspiration conditions even with the catheter in a static position. Furthermore, some portion of rapidly injected air moved from the right atrium into the right ventricle, and even with ICE image-guided catheter manipulation aspiration of intraventricular air was not possible, as the CVC used could not be advanced that far. However, it is possible to build a steerable ICE catheter with a separate lumen to allow both better visualization of intracardiac air and the targeted aspiration of intraatrial and intraventricular air, thus combining highly sensitive air detection with an improved therapeutic action.
Our study has limitations. First, we used swine but this is an established approach in VAE trials.14,49,50 Second, the exquisite sensitivity of ICE might cause false-positive results, especially with IV fluids being administered. As this was not investigated in our study, the influence of IV fluids on the correct detection of VAE using ICE needs to be evaluated in further trials. However, post hoc data analysis yielded similar results when equivocal detection episodes were counted as either positive or negative air detection.
Third, to ensure air entrainment and that injected dosages reached the heart, at least to a large extent, swine were brought into a 45° head-up position, as described previously.14 Although this is the position most often used clinically, some surgeons may perform operations in an even more upright position. However, although this may evoke a greater risk of air entrainment, it should not have affected our observations. Although the line used for air injections was perfused continuously with lactated Ringer’s solution, we cannot exclude that some portion of air injected was lodged in the vascular tree and did not reach the heart within the observation period. However, as methods of detection and bolus sizes were applied in random order, this should not have affected our results.
Investigator bias was eliminated by injecting air within a time window, with the assessor being unaware of the time of injections, and injections were performed in a randomized order to exclude dose- and time-related effects.
Although we compared the traditional aspiration technique with the new ultrasound-guided catheter displacement technique, we cannot pinpoint whether better imaging of intracardiac air, catheter manipulation, or a combination of both led to the improvement in air recovery. Furthermore, the risk of procedural-related infections remains unclear and needs to be investigated.
In conclusion, transvenous ICE is a new and exquisitely sensitive technique for the detection of VAE with much greater sensitivity than TEE and PCD techniques. Furthermore, ultrasound-guided aspiration enhanced air recovery. These encouraging results justify implementing ICE trials in patients operated upon in the sitting position.
Footnotes
Accepted for publication August 30, 2007.
Supported by Siemens, Erlangen, Germany.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
