Comparison of Calibrated and Uncalibrated Arterial Pressure Based Cardiac Output Monitors During Orthotopic Liver Transplantation

Transcription

1 LIVER TRANSPLANTATION 16: , 2010 ORIGINAL ARTICLE Comparison of Calibrated and Uncalibrated Arterial Pressure Based Cardiac Output Monitors During Orthotopic Liver Transplantation Vladimir Krejci, 1 Andrea Vannucci, 1 Alhan Abbas, 1 William Chapman, 2 and Ivan M. Kangrga 1 1 Departments of Anesthesiology; and 2 Surgery, Washington University School of Medicine, St. Louis, MO Arterial pressure based cardiac output monitors (APCOs) are increasingly used as alternatives to thermodilution. Validation of these evolving technologies in high-risk surgery is still ongoing. In liver transplantation, FloTrac-Vigileo (Edwards Lifesciences) has limited correlation with thermodilution, whereas LiDCO Plus (LiDCO Ltd.) has not been tested intraoperatively. Our goal was to directly compare the 2 proprietary APCO algorithms as alternatives to pulmonary artery catheter thermodilution in orthotopic liver transplantation (OLT). The cardiac index (CI) was measured simultaneously in 20 OLT patients at prospectively defined surgical landmarks with the LiDCO Plus monitor (CI L ) and the FloTrac-Vigileo monitor (CI V ). LiDCO Plus was calibrated according to the manufacturer s instructions. FloTrac-Vigileo did not require calibration. The reference CI was derived from pulmonary artery catheter intermittent thermodilution (CI TD ). CI V -CI TD bias ranged from 1.38 (95% confidence interval ¼ 2.02 to 0.75 L/minute/m 2, P ¼ 0.02) to 2.51 L/minute/m 2 (95% confidence interval ¼ 3.36 to 1.65 L/minute/m 2, P < 0.001), and CI L -CI TD bias ranged from 0.65 (95% confidence interval ¼ 1.29 to 0.01 L/minute/ m 2, P ¼ 0.047) to 1.48 L/minute/m 2 (95% confidence interval ¼ 2.37 to 0.60 L/minute/m 2, P < 0.01). For both APCOs, bias to CI TD was correlated with the systemic vascular resistance index, with a stronger dependence for FloTrac-Vigileo. The capability of the APCOs for tracking changes in CI TD was assessed with a 4-quadrant plot for directional changes and with receiver operating characteristic curves for specificity and sensitivity. The performance of both APCOs was poor in detecting increases and fair in detecting decreases in CI TD. In conclusion, the calibrated and uncalibrated APCOs perform differently during OLT. Although the calibrated APCO is less influenced by changes in the systemic vascular resistance, neither device can be used interchangeably with thermodilution to monitor cardiac output during liver transplantation. Liver Transpl 16: , VC 2010 AASLD. Received December 10, 2009; accepted February 15, Orthotopic liver transplantation (OLT) in patients with end-stage liver disease presents a unique hemodynamic challenge for anesthesiologists and critical care physicians. The circulatory system in cirrhosis is marked by high cardiac output, low systemic vascular resistance (SVR), tachycardia, and cardiomyopathy. 1-3 Abbreviations: ANOVA, analysis of variance; APCO, arterial pressure based cardiac output monitor; BMI, body mass index; CI, cardiac index; CI APCO, cardiac index measured with an arterial pressure based cardiac output monitor; CI L, cardiac index measured with the LiDCO Plus monitor; CI TD, cardiac index measured by thermodilution; CI V, cardiac index measured with the FloTrac-Vigileo monitor; COMP, vascular compliance; CVP, central venous pressure; EtCO 2, partial pressure of end-tidal carbon dioxide; FiO 2, fraction of inspired oxygen; Hct, hematocrit of arterial blood; HR, heart rate; INR, international normalized ratio; MAP, mean systemic arterial blood pressure; MELD, Model for End-Stage Liver Disease; MV, expiratory minute ventilation; OLT, orthotopic liver transplantation; PAC, pulmonary artery catheter; PAC-TD, pulmonary artery catheter intermittent thermodilution; PAOP, pulmonary artery occlusion pressure; PAP, mean pulmonary arterial blood pressure; pco 2, partial pressure of carbon dioxide in arterial blood; po 2, partial pressure of oxygen in arterial blood; PVRI, pulmonary vascular resistance index; ROC, receiver operating characteristic; SD, standard deviation; SVI, stroke volume index; SvO 2, mixed venous oxygen saturation; SVR, systemic vascular resistance; SVRI, systemic vascular resistance index; T eso, esophageal temperature. This study was funded by a Division of Clinical and Translational Research grant to Ivan M. Kangrga (ClinicalTrials.gov identifier: NCT ). *Current affiliation of Vladimir Krejci is Department of Anesthesiology, University Hospital of Berne, Berne, Switzerland. Address reprint requests to Andrea Vannucci, Department of Anesthesiology, Washington University School of Medicine, 660 South Euclid Avenue, Box 8054, St. Louis, MO Telephone: ; FAX: ; vannucca@wustl.edu DOI /lt Published online in Wiley InterScience ( VC 2010 American Association for the Study of Liver Diseases.

2 774 KREJCI ET AL. This underlying hyperdynamic circulation is complicated intraoperatively by frequent changes in the preload and afterload due to surgical clamping and unclamping of major abdominal vessels, hemorrhage, reperfusion, and changes in myocardial contractility. 4 This complex and dynamic circulatory physiology mandates intensive intraoperative hemodynamic monitoring. Most centers have traditionally used pulmonary artery catheter intermittent thermodilution (PAC- TD) as a standard method for monitoring cardiac output during liver transplantation and in the intensive care unit. 5 The usefulness of pulmonary artery catheters (PACs) has been surrounded by a growing controversy. The principal criticism is that the invasiveness of PACs is not supported by improved outcomes 6,7 and that, in fact, the use of PACs may be associated with a higher rate of complications and worse outcomes. 8 This controversy has prompted the development of alternative, less invasive technologies for the assessment of cardiac output, including devices that use analysis of the peripheral arterial pressure waveform [arterial pressure based cardiac output monitors (APCOs)]. Although several currently available devices share some conceptual similarities, they differ in many technological aspects, such as proprietary analysis algorithms and the need for calibration. The aim of this study was to investigate the intraoperative performance of 2 Food and Drug Administration approved and commercially available APCOs in patients undergoing OLT by comparison to the standard, PAC-TD. The 2 devices are the LiDCO Plus monitor (LiDCO Plus, Ltd., Cambridge, United Kingdom) and the FloTrac-Vigileo monitor (Edwards Lifesciences, Irvine, CA). LiDCO Plus calculates continuous cardiac output with a pulse power algorithm 9 that requires calibration by a lithium dilution cardiac output determination. In contrast, FloTrac-Vigileo uses waveform morphology-based analysis. FloTrac-Vigileo does not require calibration but rather estimates cardiac output on the basis of the patient s morphometrics and changes in vascular compliance calculated with an internal algorithm. 10 Other studies have evaluated the intraoperative performance of FloTrac-Vigileo in liver transplantation, 11,12 but LiDCO Plus has been evaluated only in postoperative liver transplant patients with stable hemodynamics. 13 The implications of this study are 2-fold. First, should LiDCO Plus, FloTrac-Vigileo, or both demonstrate good agreement with intermittent thermodilution, then these alternative technologies could potentially provide a minimally invasive and safer mode of assessment of cardiac output in this critically ill patient population. Such clinical tools would be welcomed by perioperative physicians. Second, we wondered whether a morphology-based or pulse power algorithm functions better in the setting of uneven cardiac output distribution in patients with end-stage liver disease undergoing liver transplantation. Such a direct intraoperative comparison has not been reported. PATIENTS AND METHODS Inclusion Criteria The present study was conducted in accordance with the ethical guidelines of the 1975 Declaration of Helsinki, and the protocol was approved by the institutional review board of the Washington University School of Medicine. Written, informed consent was obtained from 20 consecutive adult patients scheduled for liver transplantation who agreed to participate in the study. The data collection part of this study was conducted from July to November We excluded patients receiving lithium-based medication, patients weighing less than 40 kg, pregnant patients, and patients with acute liver failure, preexisting severe cardiac disease (including severe valvular disease), or pulmonary disease. Measurement of Cardiac Output from the Arterial Blood Pressure Waveform If not already in place, a large-bore forearm or antecubital intravenous cannula was inserted into the preoperative holding area. After premedication with a small dose (1-2 mg intravenously) of midazolam, a 20-G arterial cannula (Arrow International, Reading, PA) was inserted into the patient s radial artery under local anesthesia. The choice of the side was left to the discretion of the anesthesiologist and was based on clinical criteria and the patient s preference. The arterial cannula was connected to a FloTrac pressure transducer (FloTrac sensor, Edwards Lifesciences) with tubing provided by the manufacturer. This transducer provided 2 outputs: one to a multimodular patient monitor (Philips IntelliVue MP70, Philips Electronics North America Corp., New York, NY) for continuous arterial blood pressure display and the other to a Flo- Trac-Vigileo monitor (version 01.10, Edwards Lifesciences) for continuous cardiac index (CI) display. LiDCO Plus (LiDCO Plus, Ltd., London, United Kingdom) received the arterial pressure signal via an analogue port of the Philips monitor. Thus, all monitors received their arterial signals from the same source. The quality of the arterial transduction was assessed by continuous visual waveform inspection, which ensured line patency and square wave testing. Calibration of the LiDCO Plus Monitor The LiDCO Plus monitor was calibrated in all patients under stable hemodynamic conditions before the induction of anesthesia according to the manufacturer s instructions. Briefly, a single bolus of lithium chloride (0.3 mmol, 2 ml) was injected intravenously either with a large-bore forearm cannula 14,15 or a central venous catheter, if it was available. With each bolus injection, a lithium concentration time dilution curve was generated automatically by a lithium selective electrode connected to the tubing of the arterial line, and it was displayed on the LIDCO Plus monitor. The absolute cardiac output was then calculated from

3 CARDIAC OUTPUT MONITORS IN LIVER TRANSPLANTATION 775 the dilution curve by LiDCO Plus. At least 2 lithium dilution curves were averaged for each patient, and they were performed 5 or more minutes apart according to the manufacturer s instructions. Measurement of the Thermodilution Cardiac Output The intermittent cardiac output was measured by the injection of 10 ml of room-temperature normal saline through the venous port of an 8-Fr PAC (Opticath catheter, Hospira, Inc., Lake Forest, IL) at prospectively defined time points, as outlined later. At least 3 consecutive injections by the same anesthesia provider were performed randomly throughout the ventilatory cycle. The results were accepted and averaged if the following conditions were met: stable hemodynamic conditions and no arrhythmias throughout the measurement, a stable baseline temperature, a consistent shape of the thermodilution curve, and less than 20% variation between individual measurements and their average. For each measurement exceeding 20% variability, an additional measurement was performed, and the 3 values closest to the mean were averaged. Anesthetic Management and Monitoring After intravenous induction of general anesthesia, all patients were endotracheally intubated and mechanically ventilated (Draeger Apollo, Draeger Medical, Inc., Telford, PA). The ventilator settings were initially set according to the patient s biometric data and subsequently adjusted according to measurements of respiratory and blood gas parameters. The anesthetic management was left to the discretion of the attending anesthesiologist, who was dedicated solely to patient care and was not involved in data collection. The basis of the anesthetic management was a balanced anesthetic technique with halogenated agents, intravenous opioids, and neuromuscular blocking agents. Intravenous fluids, blood products, drugs, and interventions required for maintaining hemodynamic, metabolic, and coagulation homeostasis were administered according to clinical requirements. All patients received pre-incision antibiotic prophylaxis and 1000 mg of methylprednisolone during the anhepatic phase. Patient monitoring included 2-lead electrocardiography, pulse oximetry, noninvasive blood pressure, internal temperature, and urine output measurements. Invasive monitors measured the continuous arterial blood pressure, central venous pressure, pulmonary artery pressures, and mixed venous saturation. The intermittent thermodilution cardiac output was measured according to the study protocol and clinical indications. Transesophageal echocardiography was performed in selected patients according to clinical indications. Blood samples for blood gas, electrolyte, and coagulation analysis were drawn from the radial artery at prospectively defined time points and according to clinical necessity. Surgical Management The surgical technique was left to the discretion of the attending surgeon. A piggyback technique with a temporary portocaval shunt was used in all but 3 patients in whom surgery was performed without a portocaval shunt. Venovenous bypass was not performed in any of the patients, and no patient had bicaval graft placement. Data Collection Data sets of thermodilution-based and arterial pressure based cardiac output were obtained at predefined stages of the surgical procedure: before the induction of general anesthesia (T0; APCO and lithium dilution only), before skin incision (T1), after clamping of the portal vein (T2), after completion of the portocaval shunt (T3), before graft reperfusion (T4), after graft reperfusion (T5), and before abdominal closure (T6). The FloTrac-Vigileo and ventilator parameters were displayed on the Philips monitor via modular interfaces (Philips VueLink, Philips Electronics North America). These data were recorded continuously on a portable computer with data acquisition software (TrendFace 1.03, Ixcellence GmbH, Wildau, Germany). LiDCO Plus data (continuous cardiac output, arterial blood pressure, and heart rate) were displayed on the LiDCO Plus monitor and saved to the device s internal hard drive. The intermittent thermodilution cardiac output was acquired with a dedicated Philips MP50 monitor. Electronic event markers were used to identify the start and end of each thermodilution cardiac output measurement. Data collection with the LiDCO Plus monitor began in the preoperative holding area after insertion of the arterial cannula. The recording of all other parameters began in the operating room before the induction of general anesthesia and was discontinued when the protocol was complete. For backup and reference purposes, all electronically recorded parameters were also entered manually into a separate database. The internal clocks of all devices were synchronized before the recording was started. LiDCO Plus and Flo- Trac-Vigileo sampled the arterial waveform signal at the rate of 100 Hz. All signals were recorded as numerical values at 1 Hz. In the period surrounding the reperfusion, the recording frequency was increased to 125 Hz, and this allowed the continuous recording of physiological waveforms. Data Analysis All digitally stored data were exported as text files for analysis. All other data (eg, laboratory parameters,

4 776 KREJCI ET AL. fluids, drugs, events, and patient data) were stored in a separate database. Electronic data postprocessing was performed offline by 2 analysts. Corresponding segments of Trendface-originated and LiDCO Plus originated files were merged into a single file and further analyzed. LiDCO Plus and FloTrac-Vigileo continuous cardiac output values were averaged over the exact interval from the beginning of the first thermodilution curve to the end of the last thermodilution curve for each corresponding cardiac output measurement. Thermodilution cardiac output measurements were identified with manually recorded times of the bolus injections and LiDCO Plus monitor event markers. The exact start and stop of thermodilution injections were marked by the characteristic artifacts in the central venous pressure trace (resulting from manipulations of the 3-way stopcock of the atrial port of the PAC). Statistics The cardiac output, stroke volume, SVR, and pulmonary vascular resistance were indexed to the body surface area (m 2 ). Intraoperative hemodynamic, respiratory, and metabolic parameters were analyzed with 1-way analysis of variance (ANOVA; Prism 5.0 for Mac OS X, GraphPad Software, Inc.). The cardiac index measured by thermodilution (CI TD ) was compared with each corresponding arterial pressure waveform based CI [the cardiac index measured with the LiDCO Plus monitor (CI L ) and the cardiac index measured with the FloTrac-Vigileo monitor (CI V )] with Bland-Altman statistics (Analyse-it, Analyse-it Software, Ltd.). The difference between the bias for CI L -CI TD and CI V - CI TD and the time effect was further analyzed with 2- way ANOVA (Prism 5.0 for Mac OS X, GraphPad Software). The percentage error was calculated for each method as follows 13 : 100ð2SD of biasþ=½ðci TD þ CI APCO Þ=2Š where CI APCO is the cardiac index measured with an arterial pressure based cardiac output monitor and SD is the standard deviation. Linear correlation and regression were used to analyze the relationship between the bias and SVR. Concordance between CI TD and CI APCO was scored as the percentage of data pairs that agreed with respect to the directional change. 12 Receiver operating characteristic (ROC) analysis 14 (Prism 5.0 for Mac OS X, GraphPad Software) was used to describe the sensitivity and specificity of the arterial pressure based methods for detecting relative changes in CI TD. RESULTS The OLT procedure was completed in all 20 patients. One patient with large intrahepatic arteriovenous shunts was excluded from the analysis because of the inability to complete a satisfactory LiDCO Plus calibration. In 3 patients, no portocaval shunt was performed. A few other data points were lost for technical TABLE 1. Preinduction Parameters of Nineteen Patients Undergoing Orthotopic Liver Transplantation Age, years Female, n (%) 6 (32%) Weight, kg Height, cm BMI, kg/m Viral hepatitis, n (%) 9 (47%) Alcoholic liver disease, n (%) 2 (11%) Primary biliary cirrhosis, n (%) 2 (11%) Sclerosing cholangitis, n (%) 2 (11%) Other, n (%) 4 (21%) INR Serum creatinine, lmol/l Serum bilirubin, lmol/l MELD score HR, bpm MAP, mm Hg CI V, L/minute/m CI L, L/minute/m Hct, % NOTE: Results are presented as means and standard deviations unless otherwise indicated. reasons, so overall, 97 CI V -CI TD pairs and 107 CI L - CI TD pairs were compared. Table 1 contains demographic data and clinical characteristics from 19 patients; a few preinduction parameters are included. Intraoperative data are summarized in Tables 2 and 3. The average CI TD value was L/minute/ m 2 at the baseline, and it increased to L/ minute/m 2 (P < 0.01) at the end of the procedure. CI L increased from to L/minute/m 2, but the increase was statistically not significant. CI V remained unchanged throughout the procedure ( to L/minute/m 2 ). Bland-Altman statistical analysis was performed separately at each of the 6 predefined surgical landmarks (Fig. 1). Figure 1 shows the bias, limits of agreement, and percentage error separately for each of the 6 measurement points. With 2-way ANOVA, a significant difference was detected between CI V -CI TD and CI L -CI TD (P < 0.001) as well as a significant time effect on bias (P ¼ 0.011). The cumulative average LiDCO Plus statistics were as follows: a CI L -CI TD bias of 0.99 L/minute/m 2, upper and lower limits of agreement of 2.35 and 4.33 L/minute/m 2, and an error percentage of 75.6%. The corresponding FloTrac-Vigileo values were 1.78 L/minute/m 2 (CI V -CI TD ), 0.99 and 4.56 L/ minute/m 2, and 68.5%. Figure 2 demonstrates that the magnitude of bias for both APCOs was inversely correlated with SVR. The bias-svr relationship was second-order for both FloTrac-Vigileo and LiDCO Plus. The relationship between the cumulative bias and the Model for End-Stage Liver Disease score did not reach statistical significance for either APCO.

5 CARDIAC OUTPUT MONITORS IN LIVER TRANSPLANTATION 777 TABLE 2. Intraoperative Parameters After Induction of General Anesthesia After Clamping of Portal Vein After Opening of Temporary Portocaval Shunt Before Venous Reperfusion of Graft Liver After Venous Reperfusion of Graft Liver After Closure of Abdominal Wall P ANOVA HR, bpm MAP,mmHg PAP,mmHg <0.01 CVP,mmHg PAOP, mm Hg * CI TD, L/minute/m * <0.01 CI V, L/minute/m CI L, L/minute/m SVI, ml/m SVRI, dyn/ second/cm 5 /m 2 PVRI, dyn/ second/cm 5 /m 2 COMP, ml/ mm Hg/m 2 SvO 2,% Hct, % * <0.01 FiO 2,% EtCO 2,mmHg * MV, L/minute * * 0.01 T eso, C Arterial ph * * <0.01 pco 2,mmHg po 2,mmHg NOTE: Intraoperative parameters are presented separately for each surgical landmark. All results are presented as means and standard deviations. Statistical analysis was performed with 1-way ANOVA. P < 0.05 was considered significant. *P < 0.05 versus the corresponding measurement after induction (Dunnett s posttest). The ability of both APCOs to track changes in 2 consecutive measurements of CI TD is plotted in Fig. 3. For both monitors, the correlation with changes in CI TD was poor: R 2 was 0.20 for the FloTrac-Vigileo monitor and 0.23 for the LiDCO Plus monitor. The concordance rate for detecting the direction of cardiac output changes (increases or decreases) between CI TD and CI APCO over the whole range of consecutive measurements was 64% (50 of 78 data pairs) for FloTrac- Vigileo and 68% (53 of 78 data pairs) for LiDCO Plus (P ¼ 0.74, 95% confidence interval of the difference between the techniques ¼ 11% to 19%). Taking into account the reported variability of thermodilution measurements, for the purpose of this study, we defined a true change in the cardiac output as 620% or more variation in 2 consecutive CI TD determinations. In 41 instances, CI TD detected a true change in CI. Out of these 41 observations, FloTrac-Vigileo detected a 20% variation in CI 28 times, and LiDCO Plus detected a 20% variation in CI 30 times. Thus, the concordance between FloTrac-Vigileo and CI TD was 68%, and that between LiDCO Plus and CI TD was 73%. Using Fisher s exact test (InStat 3.0 for Mac), we did not find a significant difference between the 2 APCOs (P ¼ 0.81, 95% confidence interval of the difference between the monitors ¼ 15% to 25%). The predictive value of APCOs for detecting a 620% change in CI TD was analyzed separately for decreases and increases with ROC curves 14 (Fig. 4). The predictive value of APCOs was poor for detecting an increase in CI TD [the area under the curve was 0.66 ( , P ¼ 0.034) for CI L and 0.61 ( , P ¼ 0.12) for CI V ] and fair for detecting a decrease in CI TD [the area under the curve was 0.79 ( , P ¼ 0.001) for CI L and 0.73 ( , P ¼ ) for CI V ]. DISCUSSION Our study is the first to evaluate the intraoperative performance of the LiDCO Plus monitor during liver transplantation. LiDCO Plus calculates beat-to-beat cardiac output by combining an arterial waveform analysis algorithm (PulseCO) with a single-point lithium indicator dilution calibration (LiDCO Plus). 9,16 This method has shown good agreement with thermodilution in different patient populations but has not been tested as an intraoperative monitor in patients undergoing OLT. In fact, LiDCO Plus has

6 778 KREJCI ET AL. TABLE 3. Summary of Surgical and Anesthetic Data Length of surgery, minutes 498 ( ) Portocaval shunt, n (%) 16 (84%) Blood loss, ml 3828 (750-15,000) Urine output, ml 534 ( ) Crystalloids, ml 7395 ( ,000) Colloids, ml 1360 ( ) Packed red blood cells, U 6.5 (0-26) Fresh frozen plasma, U 4.1 (0-22) Platelets, U 0.5 (0-3) Cryoprecipitate, U 4.7 (0-20) Time from calibration to 281 ( ) reperfusion (min) Time from calibration to last 438 ( ) measurement (min) NOTE: Results are presented as means and ranges unless otherwise indicated. only once been compared with thermodilution in spontaneously breathing, post liver transplant patients in the intensive care unit in whom high cardiac output was associated with stable hemodynamics. 18 The favorable results from this previous study may not be fully applicable to the operating room setting. Our intraoperative comparison of LiDCO Plus and PAC-TD (Fig. 1) demonstrates consistently, over all measurement points, larger bias and limits of agreement than previously reported for posttransplant patients. The error percentage in our study was above the accepted limit of 30% 19 at all measurement points. Our results suggest that the LiDCO Plus monitor cannot be used interchangeably with PAC-TD to assess CI in OLT. LiDCO Plus bias, as shown in Fig. 2, increased over time and was also slightly influenced by SVR. Furthermore, changes in CI L only partially reflected changes in PAC-TD CI throughout the whole procedure, as shown in Fig. 3. Figure 1. Comparison of CI TD and CIs measured with arterial pressure analysis during OLT. Open circles compare CI TD with CI V ; gray circles compare CI TD with CI L. Bland-Altman statistics and the percentage error were calculated separately for each surgical landmark.

7 CARDIAC OUTPUT MONITORS IN LIVER TRANSPLANTATION 779 Figure 2. Relationship between the bias and systemic vascular resistance index (SVRI): (A) Vigileo and (B) LiDCO. Figure 3. Relationship between the percentage changes in CI of 2 consecutive measurements. Black circles compare CI TD and CI V ; gray circles compare CI TD and CI L. In a clinical setting, both the absolute value and the relative changes in CI guide clinician s decisions. We assessed the capability of LiDCO Plus for detecting changes in CI measured by PAC-TD. We used a 20% PAC-TD change as the cutoff as this value is larger than the usually recognized measurement variability and is likely to represent a true change in CI. The overall capability of LiDCO Plus for assessing PAC-TD CI changes greater than 20% was poor for increases and fair for decreases (Fig. 4). 20 Although we do not have an exact explanation for this difference, this observation is consistent with a hypothesis of dual circulation with preferential distribution of CI increases toward the visceral compartment, as suggested by Newby and Hayes 21 and corroborated by Niemann et al. 22 (discussed later). In our study, only 1 LiDCO Plus calibration was performed, as the manufacturer recommends 1 calibration every 24 hours. Calibration was performed prior to induction to ensure stable hemodynamic conditions and avoid interference by neuromuscular blockers. More recent evidence suggests that the agreement between the lithium dilution cardiac output and pulse power algorithm in critically ill patients may be acceptable for no longer than 4 hours. 23 In our study, LiDCO Plus PAC-TD bias increased with time. More frequent calibration could have affected the absolute values of the LiDCO Plus cardiac output

8 780 KREJCI ET AL. Figure 4. ROC curves plotting the sensitivity versus specificity in order to evaluate the ability of arterial pressure based CIs to predict a 20% change in CI TD. Curves were plotted separately for increases and decreases. Open circles indicate changes in CI V ; gray circles indicate changes in CI L. and hence the bias, but the tracking capability of LiDCO Plus, evaluated as the magnitude and direction of percentage CI changes, would have been the same. FloTrac-Vigileo is easy to set up and requires no calibration. It calculates cardiac output only on the basis of an analysis of the peripheral arterial pressure waveform and patient demographic data. 10 The calculated cardiac output is updated every 20 seconds. Flo- Trac-Vigileo has been tested in operative and intensive care unit settings, and conflicting results have been reported In one study in liver transplant patients, Biais et al. 11 studied the agreement between FloTrac-Vigileo and a STAT-Mode automatic thermodilution PAC, which determines cardiac output over 60 seconds. An acceptable agreement was demonstrated only in a subset of patients with relatively preserved SVR (Child-Pugh A) and not in those with lower SVR (Child-Pugh B and C). Similar observations were made in a more recent study. 12 In patients up to 48 hours post liver transplant, Della Rocca and colleagues 27 reported acceptable agreement of FloTrac- Vigileo and a continuous cardiac output PAC in a group of patients with cardiac outputs below 8 L/minute but not in patients with outputs above 8 L/minute. These studies suggest that FloTrac-Vigileo does not perform well in the rapidly changing hemodynamics of liver transplantation, particularly under hyperdynamic conditions. Our results, demonstrating a large bias and unacceptable error percentage (Fig. 1), a strong dependence of bias on SVR (Fig. 2), and poor tracking capability (Figs. 3 and 4), are in agreement with the previous studies. Of interest is the observation of a consistent underestimation of PAC-TD by FloTrac-Vigileo. An updated version of the FloTrac- Vigileo algorithm (third generation) has recently been released. It remains to be seen whether some of the described limitations have been overcome. We recognize several limitations of the current study. First, the PAC was placed only after the induction of general anesthesia, whereas LiDCO Plus calibration was performed before the induction. As a result, we cannot compare preinduction APCO and CI TD values. Second, all APCO measurements were based on the radial artery pressure waveform. Under conditions of extreme splanchnic vasodilation and systemic hypotension with concomitant vasopressor administration, a peripheral artery site may not accurately reflect aortic pressure. It is possible that recording from the femoral artery would have yielded different results. 28 It should be noted that the manufacturers of both devices state that a peripheral arterial line is adequate, and thus using only a radial arterial line has become common clinical practice. 29,30 It has been proposed that in hyperdynamic patients with cirrhosis, cardiac output is distributed to multiple parallel circulations. 2,21 These vascular beds can be hyperperfused (splanchnic), normoperfused, or hypoperfused (kidney). Consistent with this hypothesis, Niemann et al., 22 using a model based on indocyanine green pharmacokinetics, demonstrated that patients undergoing liver transplantation have increased distribution of cardiac output in the central and splanchnic compartment in comparison with healthy living liver donors. As this study was limited to the dissection phase, it is not known whether and how the overall cardiac output is redistributed between the different vascular beds during the anhepatic phase and after reperfusion. The APCOs base their function on the assumption that the cardiac output changes are distributed evenly, and there is no evidence for this in liver transplantation. It is then possible to speculate that cardiac output as assessed by a peripheral artery waveform, independently of the effects of vasoactive drugs, cannot reflect global cardiac output changes as assessed by thermodilution. The accuracy and reproducibility of the thermodilution cardiac output could have been higher if an icecold injectate had been used. However, studies using

9 CARDIAC OUTPUT MONITORS IN LIVER TRANSPLANTATION 781 at least 3 measurements of room-temperature injectate (10 ml) have shown comparable reproducibility with an iced indicator. 31,32 Room-temperature injectate was used in our study to avoid hypothermia according to our clinical standard. The main point of interest of this study is that it is, to the best of our knowledge, the first intraoperative study of LiDCO Plus in liver transplantation and the first direct comparison of calibrated and uncalibrated APCOs in such a setting. The strength of this study is the real-time recording of a complete intraoperative monitoring data set sampled at a high frequency. Our setup allowed us to accurately identify and compare off-line the corresponding continuous data sets acquired by the 2 APCOs with the intermittent cardiac output measurements obtained with PAC-TD. Our data suggest that the tested calibrated and uncalibrated APCOs cannot be used interchangeably with PAC-TD to measure cardiac output in liver transplantation. REFERENCES 1. Henriksen JH, Fuglsang S, Bendtsen F, Moller S. Arterial hypertension in cirrhosis: arterial compliance, volume distribution, and central haemodynamics. Gut 2006;55: Moller S, Henriksen JH. Cardiopulmonary complications in chronic liver disease. World J Gastroenterol 2006;12: De Wolf A. Hemodynamic monitoring during orthotopic liver transplantation. Transplant Proc 1993;25: De Wolf AM. Perioperative assessment of the cardiovascular system in ESLD and transplantation. Int Anesthesiol Clin 2006;44: De Wolf AM. Pulmonary artery catheter: rest in peace? Not just quite yet. Liver Transpl 2008;14: Sandham JD, Hull RD, Brant RF, Knox L, Pineo GF, Doig CJ, et al. A randomized, controlled trial of the use of pulmonary-artery catheters in high-risk surgical patients. N Engl J Med 2003;348: Wheeler AP, Bernard GR, Thompson BT, Schoenfeld D, Wiedemann HP, deboisblanc B, et al. Pulmonary-artery versus central venous catheter to guide treatment of acute lung injury. N Engl J Med 2006;354: Connors AF Jr, Speroff T, Dawson NV, Thomas C, Harrell FE Jr, Wagner D, et al. The effectiveness of right heart catheterization in the initial care of critically ill patients. JAMA 1996;276: Rhodes A, Sunderland R. Arterial pulse power analysis: the LiDCO TM plus system. In: Pinsky MR, Payen D, eds. Functional Hemodynamic Monitoring. Berlin, Germany: Springer Verlag; 2005: Update in Intensive Care and Emergency Medicine Manecke GR Jr, Auger WR. Cardiac output determination from the arterial pressure wave: clinical testing of a novel algorithm that does not require calibration. J Cardiothorac Vasc Anesth 2007;21: Biais M, Nouette-Gaulain K, Cottenceau V, Vallet A, Cochard JF, Revel P, Sztark F. Cardiac output measurement in patients undergoing liver transplantation: pulmonary artery catheter versus uncalibrated arterial pressure waveform analysis. Anesth Analg 2008;106: Biancofiore G, Critchley LA, Lee A, Bindi L, Bisa M, Esposito M, et al. Evaluation of an uncalibrated arterial pulse contour cardiac output monitoring system in cirrhotic patients undergoing liver surgery. Br J Anaesth 2009;102: Costa MG, Della Rocca G, Chiarandini P, Mattelig S, Pompei L, Barriga MS, et al. Continuous and intermittent cardiac output measurement in hyperdynamic conditions: pulmonary artery catheter vs. lithium dilution technique. Intensive Care Med 2008;34: Garcia-Rodriguez C, Pittman J, Cassell CH, Sum-Ping J, El-Moalem H, Young C, Mark JB. Lithium dilution cardiac output measurement: a clinical assessment of central venous and peripheral venous indicator injection. Crit Care Med 2002;30: Jonas MM, Kelly FE, Linton RA, Band DM, O Brien TK, Linton NW. A comparison of lithium dilution cardiac output measurements made using central and antecubital venous injection of lithium chloride. J Clin Monit Comput 1999;15: Linton RA, Band DM, Haire KM. A new method of measuring cardiac output in man using lithium dilution. Br J Anaesth 1993;71: Linton R, Band D, O Brien T, Jonas M, Leach R. Lithium dilution cardiac output measurement: a comparison with thermodilution. Crit Care Med 1997;25: Kurita T, Morita K, Kato S, Kikura M, Horie M, Ikeda K. Comparison of the accuracy of the lithium dilution technique with the thermodilution technique for measurement of cardiac output. Br J Anaesth 1997;79: Critchley LA, Critchley JA. A meta-analysis of studies using bias and precision statistics to compare cardiac output measurement techniques. J Clin Monit Comput 1999;15: Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143: Newby DE, Hayes PC. Hyperdynamic circulation in liver cirrhosis: not peripheral vasodilatation but splanchnic steal. QJM 2002;95: Niemann CU, Yost CS, Mandell S, Henthorn TK. Evaluation of the splanchnic circulation with indocyanine green pharmacokinetics in liver transplant patients. Liver Transpl 2002;8: Cecconi M, Fawcett J, Grounds RM, Rhodes A. A prospective study to evaluate the accuracy of pulse power analysis to monitor cardiac output in critically ill patients. BMC Anesthesiol 2008;8: Meyer J, Boldt J, Poland R, Peterson A, Manecke GR. Continuous arterial pressure waveform-based cardiac output using the Flo-Trac/FloTrac-Vigileo: a review and metaanalysis. J Cardiothorac Vasc Anesth 2009;23: Button D, Weibel L, Reuthebuch O, Genoni M, Zollinger A, Hofer CK. Clinical evaluation of the FloTrac/FloTrac- Vigileo system and two established continuous cardiac output monitoring devices in patients undergoing cardiac surgery. Br J Anaesth 2007;99: Prasser C, Trabold B, Schwab A, Keyl C, Ziegler S, Wiesenack C. Evaluation of an improved algorithm for arterial pressure-based cardiac output assessment without external calibration. Intensive Care Med 2007;33: Della Rocca G, Costa MG, Chiarandini P, Bertossi G, Lugano M, Pompei L, et al. Arterial pulse cardiac output agreement with thermodilution in patients in hyperdynamic conditions. J Cardiothorac Vasc Anesth 2008;22: Shin BS, Kim GS, Ko JS, Gwak MS, Yang CS, Hahm TS, Lee S-K. Comparison of femoral arterial blood pressure with radial arterial blood pressure and noninvasive

10 782 KREJCI ET AL. upper arm blood pressure in the reperfusion period during liver transplantation. Transplant Proc 2007;39: Rhodes A, Grounds RM. New technologies for measuring cardiac output: the future? Curr Opin Crit Care 2005; 11: Linton NW, Linton RA. Estimation of changes in cardiac output from the arterial blood pressure waveform in the upper limb. Br J Anaesth 2001;86: Elkayam U, Berkley R, Azen S, Weber L, Geva B, Henry WL. Cardiac output by thermodilution technique. Effect of injectate s volume and temperature on accuracy and reproducibility in the critically ill patient. Chest 1983;84: Bazaral MG, Petre J, Novoa R. Errors in thermodilution cardiac output measurements caused by rapid pulmonary artery temperature decreases after cardiopulmonary bypass. Anesthesiology 1992;77:31-33.