- Top of page
The aim of this study was to compare the accuracy of pulse dye densitometry with that of bolus thermodilution cardiac output measurement in patients before and after elective coronary artery bypass grafting. Twenty-eight patients were studied. Agreement between mean thermodilution and pulse dye densitometry cardiac output values was assessed by Bland-Altman analysis. Preoperative median [range] cardiac output was 3.87 [2.37–6.0] l.min−1 by thermodilution, and 3.11 [1.7–5.45] l.min−1 by pulse dye densitometry using indocyanine green 5 mg. Pulse dye densitometry underestimated cardiac output (mean bias − 0.42 l.min−1); the limits of agreement were ± 1.91 l.min−1, and mean error was 50.3%, indicating low precision. Preoperative median [range] cardiac output was 3.85 [2.2–6.0] l.min−1 for bolus thermodilution cardiac output and 4.2 [2.0–7.2] l.min−1 for pulse dye densitometry using indocyanine green 20 mg. Mean bias was + 0.566 l.min−1, the limits of agreement were ± 2.51 l.min−1 and mean error was 60.9%. Postoperative cardiac output data were not analysed because pulse dye densitometry signals were low or absent in > 50% of the patients. We conclude that pulse dye densitometry using indocyanine green 5 mg or 20 mg is inaccurate in anaesthetised patients before coronary artery bypass surgery and cannot be used after surgery because of a high incidence of low pulse dye densitometry signal amplitudes.
Attempts have been made in the last few decades to develop non-invasive or semi-invasive methods of measuring cardiac output, such as transthoracic bio-impedance , whole body bio-impedance  and Doppler techniques . More recently, pulse dye densitometry has been proposed as a viable alternative to semi-invasive cardiac output determination [4–6]. Pulse dye densitometry is based on the principle of indocyanine green dye dilution, whereby indocyanine green is injected through a central  or peripheral  venous line and the changes in dye concentration over time are detected by pulse spectrophotometry using a finger probe [7, 9] or a nose probe [10, 11]. Only a few investigators [7, 12–14] have compared pulse dye densitometry with standard methods of cardiac output determination, and the accuracy of pulse dye densitometry is therefore in dispute.
Our primary aim was to validate pulse dye densitometry in patients before and after elective coronary artery bypass graft surgery (CABG). Our secondary aim was to investigate whether different concentrations of indocyanine green would affect the accuracy of pulse dye densitometry. The accuracy of cardiac output determination by pulmonary thermodilution has been assessed and confirmed at our institution .
- Top of page
With the approval of the Local Research Ethics Committee and written, informed consent, 28 patients who were scheduled for elective CABG were enrolled in the study. Exclusion criteria comprised: allergy to iodine, presence of an aortic counterpulsation device, severe renal disease or liver dysfunction and intake of drugs with a possible influence on indocyanine green uptake [16–18]. Patients were premedicated with oral flunitrazepam 1–2 mg (0.015–0.03 mg.kg−1) the evening before surgery and oral midazolam 0.1 mg.kg−1 45 min before transfer to the operating theatre. On arrival, arterial and venous cannulae were inserted and both pulse oximetry and five-lead electrocardiographic monitoring were started. Anaesthesia was induced with midazolam 30–100 μg.kg−1, fentanyl 3–10 μg.kg−1 and pancuronium 0.1 mg.kg−1. After tracheal intubation, the patients' lungs were mechanically ventilated (Siemens Servo 900, Erlangen, Germany). A triple-lumen central venous catheter (Arrow International, Reading, PA) and an 8.5 FG introducer (Arrow International) were inserted via the right internal jugular vein. A 7.5 FG thermistor-tipped, flow-directed pulmonary artery catheter (IntelliCath, Baxter Healthcare Corporation, Edwards Critical Care Division, Irvine, CA) was then introduced. Anaesthesia was maintained with continuous infusions of propofol 2–6 mg.kg−1.h−1, remifentanil 0.1–0.3 μg.kg−1.min−1 and bolus injections of fentanyl before skin incision, sternotomy and chest closure.
Thermodilution cardiac output was measured with the pulmonary artery catheter connected to a cardiac output measuring device (9520A, Baxter Healthcare Corporation). Ten millilitres of iced saline 0.9% were injected with a closed injectate system (CO-set; Baxter Healthcare Corporation). Injections were distributed randomly throughout the respiratory cycle. Thermodilution curves were displayed on a recorder and accepted as correct if the shape of the curve fulfilled the criteria of Levett and Replogle and if the injectate temperature was < 10 °C . Between three and five thermodilution cardiac output measurements were carried out for every patient, and the mean cardiac output value was used for comparison. Thermodilution cardiac output measurements were arbitrarily defined as being acceptable if they were within 10% of one another.
Dye dilution measurements were carried out using bolus injections of indocyanine green (ICG-Pulsion, Munich, Germany). After a bolus injection, indocyanine green distributes exclusively in the intravascular space and is cleared by hepatic elimination with a half-life of 4.1 min . Since total elimination of indocyanine green would take about three half-lives, the remaining plasma indocyanine green would cause errors if measurements were repeated every 2 min during cardiac output measurements. The device therefore automatically calibrates to zero indocyanine green plasma concentration at the time a new measurement starts. This baseline level gradually decreases during the measurement as a result of the elimination of the remaining indocyanine green. Hence, the area under the indocyanine green concentration curve is underestimated but the resulting error is negligible, as previously reported . The blood concentration of indocyanine green was determined with a pulse dye densitogram analyser (DDG-2001, A/K, Nihon-Kohden, Tokyo, Japan), which was connected to a clip device attached to the patient's left forefinger.
Pulse dye densitometry uses the wavelengths 805 nm and 940 nm to calculate the ratio between indocyanine green and haemoglobin. Over a period of 30 s, the relative indocyanine green concentration is detected. After entering the haemoglobin concentration obtained by blood gas analysis (ABL 625, Radiometer, Copenhagen, Denmark), the dye densitometer calculates the absolute indocyanine green concentration according to the measured ratio of haemoglobin. A dye densitogram is displayed on the computer screen. The area under the first circulation curve is inversely proportional to blood flow, and the cardiac output is calculated from the Steward-Hamilton equation. A more detailed description is given elsewhere [2, 15]. Two concentrations of indocyanine green (5 mg and 20 mg dissolved in glucose 5% 2 ml) were used for cardiac output estimation. Indocyanine green was injected through the distal port of the central venous line, which was then flushed with glucose 5% 5 ml. The mean cardiac output value obtained after two consecutive administrations of indocyanine green 5 mg was used for comparison with cardiac output measured by thermodilution. With the higher concentration of 20 mg of indocyanine green, only one injection was administered. To obtain reliable measurements, the pulsatile component of light absorption in pulse oximetry should account for 1–5% of the total signal. Therefore, the pulse amplitude was displayed with a quality bar divided into five divisions. According to the manufacturer's specification, pulse dye densitometry measurements were not carried out when the pulse amplitude was < 1%.
Measurements were carried out under haemodynamic steady state conditions after insertion of the central venous line and the pulmonary artery catheter both before surgery and within the first 24 h after admission of the patient to the intensive care unit. Before each measurement series, blood was drawn for analysis of arterial blood gases, haemoglobin concentration and mixed venous oxygen saturation. Thereafter, standard haemodynamic parameters (heart rate and mean arterial, central venous, pulmonary artery and pulmonary capillary occlusion pressures) were documented. Three to five bolus injections of iced saline 0.9% 10 ml were then given, followed immediately by duplicate pulse dye densitometry determinations using indocyanine green 5 mg. A second blood gas analysis was then performed, and standard haemodynamic parameters were again documented. Thermodilution cardiac output measurements were repeated, and indocyanine green 20 mg was injected immediately thereafter to measure cardiac output by pulse dye densitometry. The whole series of measurements was repeated between 2 h and 24 h after intensive care unit admission.
A power calculation was performed to determine the sample size. It revealed that a sample size of 30 patients would have an 80% power to detect a difference in means of 0.5 l.min−1, assuming a standard deviation of differences of 0.8 l.min−1, using a paired t-test with a 0.05 two-sided significance level. Agreement between mean baseline cardiac output and pulse dye densitometry values was assessed by Bland-Altman analysis [22–24].
- Top of page
Twenty-eight patients were enrolled in the study, but one patient was excluded before surgery because the pulse dye densitometry device indicated a low pulse amplitude. The characteristics of the 27 patients studied are listed in Table 1. After injection of indocyanine green 20 mg, four patients had low signal amplitude or missing signals. Therefore, the pre-operative data for pulse dye densitometry using indocyanine green 20 mg were analysed in 23 patients (Table 2).
Table 1. Patient characteristics. Values are number or median [range].
|Number of subjects||27|
|Sex; M: F||23 : 4|
|Age; years||63 [49–81]|
|Weight; kg||85 [56–135]|
|Body mass index; kg.m−2||28.7 [22.2–43.1]|
|Body surface area; m2|| 1.99 [1.56–2.5]|
Table 2. Agreement between cardiac output values measured by thermodilution and pulse dye densitometry. Values are median (IQR [range]), number or percent.
| ||Indocyanine green 5 mg n = 27||Indocyanine green 20 mg n = 23|
|Thermodilution cardiac output; l.min−1|| 3.87 (3.3–4.2 [2.4–6.0])|| 3.85 (3.2–4.3 [2.2–6])|
|Pulse dye densitometry cardiac output; l.min−1|| 3.11 (2.4–4.2 [1.7–5.45])|| 4.2 (3.4–5.5 [2.0–7.2])|
|2 SD; l.min−1|| 1.9|| 2.5|
|Mean error; %||50.3||60.9|
Median (IQR [range]) cardiac output values for baseline cardiac output were 3.87 (3.3–4.2 [2.4–6.0]) l.min−1 before injection of indocyanine green 5 mg, and 3.85 (3.2–4.3 [2.2–6.0]) l.min−1 before injection of indocyanine green 20 mg. Median (IQR [range]) cardiac output values for pulse dye densitometry were 3.11 (2.4–4.2 [1.7–5.45]) l.min−1 using indocyanine green 5 mg and, 4.2 (3.4–5.2 [2.0–7.2]) l.min−1 using indocyanine green 20 mg. The two sets of baseline cardiac output values obtained before pulse dye densitometry using indocyanine green 5 mg or 20 mg were not significantly different, with a bias of − 0.08 l.min−1 and limits of agreement of ± 0.55 l.min−1, confirming steady-state conditions.
Before surgery, the mean (SD) bias was − 0.42 (0.95) l.min−1 when indocyanine green 5 mg was injected. The limits of agreement (2 SD) were 1.91 l.min−1, and the mean error 50.3%(Fig. 1). After injection of indocyanine green 20 mg, the mean (SD) bias was + 0.57 (1.25) l.min−1, the limits of agreement ± 2.5 l.min−1(Fig. 2) and the mean error 60.9% (Table 2).
Figure 1. Agreement between bolus thermodilution (BCO) and pulse dye densitometry (CO-PDD) cardiac output measurement using an injection of indocyanine green 5 mg.
Download figure to PowerPoint
Figure 2. Agreement between bolus thermodilution (BCO) and pulse dye densitometry (CO-PDD) cardiac output measurement using an injection of indocyanine green 20 mg.
Download figure to PowerPoint
Pulse dye densitometry using indocyanine green 5 mg thus underestimated cardiac output, whereas pulse dye densitometry using indocyanine green 20 mg overestimated cardiac output. Limits of agreement were high, and the mean errors were well above the acceptable limit of 30%, indicating low precision, irrespective of whether 5 mg or 20 mg of indocyanine green was injected. Contrary to the expectation of achieving better precision at higher concentrations, the mean error was higher when 20 mg of indocyanine green was injected than when 5 mg was injected (60.9% and 50.3%, respectively).
In the two pulse dye densitometry measurements with indocyanine green 5 mg, reproducibility was poor, with limits of agreement of ± 1.25 l.min−1 and a mean error of 37.3%. Pulse dye densitometry data obtained in the ICU after surgery were not analysed because pulse signals were low or absent in > 50% of the 27 patients.
- Top of page
In this study, performance of the pulse dye densitometry technique was poor in patients before and after elective CABG surgery when thermodilution cardiac output measurement was used as a reference technique. Before surgery, cardiac output was underestimated by pulse dye densitometry using indocyanine green 5 mg, and was overestimated by pulse dye densitometry using indocyanine green 20 mg. Furthermore, precision was unacceptably low irrespective of dye dose. Performance was particularly poor in the ICU, where the signals were low or absent in > 50% of the patients.
Conflicting reports have been published on the accuracy of pulse dye densitometry [7, 9, 11–14, 21]. Bühlmann et al.  and Imai et al.  reported a satisfactory agreement between pulse dye densitometry and thermodilution, but the results of the Bland-Altman analysis do not justify their conclusion. Although the mean bias was close to zero, the limits of agreement were ± 1.6 l.min−1and ± 2.7 l.min−1, respectively. Our estimates of the mean error from these data are 36% and 60%.
Imai et al. assessed the accuracy and precision of pulse dye densitometry in a heterogeneous patient population by comparing single cardiac output values obtained by simultaneous injection of both dye and thermal indicator . They reported a slight cardiac output overestimate by pulse dye densitometry (bias: + 0.16 l.min−1) and a precision of ± 0.8 l.min−1. Increased overestimation occurred, particularly in the low cardiac output range (cardiac output < 3.5 l.min−1) with a bias between dye and thermodilution of + 0.29 l.min−1 and a precision of ± 1.7 l.min−1. These results could not be confirmed by other investigators who compared pulse dye densitometry with thermodilution cardiac output in cardiac surgery patients after admission to the intensive care unit using the simultaneous injection method [9, 11, 13].
In contrast to our study, Imai et al. used a nose probe to detect indocyanine green concentrations . Fukuda et al., injecting a bolus of 20 mg of indocyanine green via a peripheral venous route, reported greater accuracy of pulse dye densitometry with a nose probe than with a finger probe, which they attributed to the decreased mean transit time for indocyanine green detection with the nose probe . Other investigators found no difference between the two sites of indocyanine green detection [9, 11]. In our study, different sites for detecting indocyanine green concentrations were not systematically examined, but we found no improvement in signal when the finger probe was changed to a nose probe in patients presenting with low signal amplitudes or missing signals.
The best agreement between pulse dye densitometry and thermodilution was reported by Bremer et al. , but the technique was only found to be reliable at cardiac output values > 5.0 l.min−1. In the subgroup of patients presenting with low cardiac output (< 5.0 l.min−1), pulse dye densitometry underestimated cardiac output values measured by thermodilution and the Fick method, and precision was low. When pulse dye densitometry was compared with cardiac output values < 5.0 l.min−1, the bias was − 0.51 l.min−1, and the limits of agreement were ± 0.8 l.min−1. Comparing pulse dye densitometry with the Fick method revealed a bias of − 0.83 l.min−1 and limits of agreement of ± 1.9 l.min−1. The corresponding values for bias and precision at cardiac output > 5.0 l.min−1 were − 0.29 l.min−1 and ± 1.14 l.min−1 (pulse dye densitometry vs. baseline cardiac output) and − 0.58 l.min−1 and ± 1.42 l.min−1 (pulse dye densitometry vs. Fick), respectively. Thermodilution cardiac output values in our patient population ranged from 2.04 to 7.24 l.min−1, and our results correspond nicely with those published by Bremer et al.  for cardiac output values < 5 l.min−1.
Some investigators have used indocyanine green concentrations ranging from 5 to 30 mg per bolus injection to assess the accuracy and precision of pulse dye densitometry in comparison with invasive methods. Sakka et al. compared pulse dye densitometry using a nose probe and injections of indocyanine green 30 mg with a transpulmonary indicator dilution technique in critically ill patients; agreement between the two methods was poor, with a bias of − 0.80 l.min−1 and limits of agreement of ± 3.4 l.min−1. To the best of our knowledge, only Sha et al.  hypothesised that the accuracy of pulse dye densitometry could depend on the indocyanine green concentration. By injecting a 20-mg indocyanine green bolus via a peripheral venous line, they found good agreement between pulse dye densitometry and thermodilution. In our study, pulse dye densitometry with indocyanine green 20 mg overestimated cardiac output. Surprisingly, pulse dye densitometry was more precise when a 5-mg bolus of indocyanine green was used.
In agreement with other investigators, we found a high incidence of low or missing pulse dye densitometry signal amplitude, particularly after surgery [9, 11, 13]. In our study, > 50% of the patients showed low signal amplitudes or absent signals after CABG surgery. This result was independent of the cardiac output value measured by thermodilution cardiac output. All patients had a positive fluid balance of 500–2000 ml during surgery. Whether the failure to detect indocyanine green with pulse spectrophotometry can be explained by changes in intercellular and intracellular fluid or poor peripheral circulation remains speculative. The pulse dye dilution technique is subject to the same limitations as pulse oximetry. Therefore, motion artifacts and direct daylight may influence signal registration. Haruna et al. compared pulse dye densitometry using indocyanine green 10 mg with a 131iodine-labelled human serum method for blood volume determination in healthy volunteers. They reported that in 30% of the cases, pulse dye densitometry measurement failed because of motion artifacts or low signal-to-noise ratio . In the same study, pulse dye densitometry measurements failed in > 30% of patients after cardiac surgery because of motion distortion or insufficient pulse amplitude. In the study by Bremer et al. , pulse dye densitometry using a nasal wing clip failed in 10% of cardiac surgery patients after admission to the intensive care unit. In concordance with our results, Hofer et al.  reported a 33% incidence of missing signals in patients after CABG surgery using a finger clip.
It is well known that solutions containing sodium sulphide, particularly in combination with heparin, can decrease the clearance of indocyanine green , but this alteration of clearance has no relevance in cardiac output determination using indocyanine green [25–27]. The influence of haemodilution on the pharmacokinetics of indocyanine green has not been thoroughly investigated. After intravenous injection, indocyanine green binds very quickly and almost completely to globulins, particularly to alpha1-lipoproteins. This complete binding prevents any uptake by peripheral tissues, kidney or lungs . Haemodilution might decrease the intravascular content of alpha1-lipoproteins. Under these circumstances, some of the indocyanine green could become unbound and disappear from the intravascular space. The amount of indocyanine green injected would thus not correspond with the amount detected, and the decreased area under the curve would produce an overestimation of cardiac output. Interestingly, Shrewsbury et al.  found a decreased indocyanine green clearance during haemodilution with Fluosol-DA or normal saline in the presence of unchanged liver blood flow in rats. They hypothesised that indocyanine green clearance was altered due to changes in its extraction ratio . However, there have been no further investigations of the influence of haemodilution on indocyanine green clearance and on the detection of indocyanine green concentrations using a finger probe or a nose probe.
Less invasive, simpler and more rapid methods of cardiac output monitoring would be highly desirable. The use of pulse dye densitometry has been hailed by some investigators as a promising new approach. In the present study, conducted in anaesthetised patients before and after CABG, pre-operative agreement between pulse dye densitometry and baseline cardiac output was poor. In agreement with other investigators, pulse dye densitometry using indocyanine green 5 mg underestimated cardiac output in low to normal cardiac output states, and pulse dye densitometry using indocyanine green 20 mg overestimated cardiac output. The precision of pulse dye densitometry measurements was low with both concentrations of indocyanine green. In the first 24 h after CABG surgery, pulse dye densitometry could not be used because of a high incidence of low signal amplitude or missing signals. Pulse dye densitometry cannot therefore be recommended for cardiac output monitoring in patients before and after CABG surgery.