- Email: [email protected]
Electrical Impedance Tomography in the Assessment of Extravascular Lung Water in Noncardiogenic Acute Respiratory Failure
Electrical Impedance Tomography in the Assessment of Extravascular Lung Water in Noncardiogenic Acute Respiratory Failure* Peter W. A. Kunst, MD; Anton Vonk Noordegraaf, MD; Esther Raaijmakers, MSC; Jan Bakker, MD, PhD; A. B. Johan Groeneveld, MD, PhD; Piet E. Postmus, MD, PhD, FCCP; and Peter M. J. M. de Vries, MD, PhD
Study objectives: To establish the value of electrical impedance tomography (EIT) in assessing pulmonary edema in noncardiogenic acute respiratory failure (ARF), as compared to the thermal dye double indicator dilution technique (TDD). Design: Prospective clinical study. Setting: ICU of a general hospital. Patients: Fourteen ARF patients. Interventions: In order to use the TDD to determine the amount of extravascular lung water (EVLW), a fiberoptic catheter was placed in the femoral artery. Measurements and main results: Fourteen consecutive ARF patients receiving mechanical ventilation were measured by EIT and TDD. EIT visualizes the impedance changes caused by the ventilation in two-dimensional image planes. An impedance ratio (IR) of the ventilation-induced impedance changes of a posterior and an anterior part of the lungs was used to indicate the amount of EVLW. For the 29 measurements in 14 patients, a significant correlation between EIT and TDD (r 5 0.85; p < 0.001) was found. The EIT reproducibility was good. The diagnostic value of the method was tested by receiver operator characteristic analysis, with 10 mL/kg of EVLW considered as the upper limit of normal. At a cutoff level of the IR of 0.64, the IR had a sensitivity of 93%, a specificity of 87%, and a positive predictive value of 87% for a supranormal amount of EVLW. Follow-up measurements were performed in 11 patients. A significant correlation was found between the changes in EVLW measured with EIT and TDD (r 5 0.85; p < 0.005). Conclusion: We conclude that EIT is a noninvasive technique for reasonably estimating the amount of EVLW in noncardiogenic ARF. (CHEST 1999; 116:1695–1702) Key words: ARDS; electrical impedance; extravascular lung water; thermal dye dilution; tomography Abbreviations: ALI 5 acute lung injury; ARF 5 acute respiratory failure; CV 5 coefficient of variation; EIT 5 electrical impedance tomography; EVLW 5 extravascular lung water; Fio2 5 fraction of inspired oxygen; ICG 5 indocyanine green; IR 5 impedance ratio; LIS 5 lung injury score; PEEP 5 positive end-expiratory pressure; RC 5 reliability coefficient; ROC 5 receiver operator characteristic; ROI 5 region of interest; TDD 5 thermal dye double indicator dilution technique; Vt 5 tidal volume
many cases of acute respiratory failure (ARF), I npulmonary edema contributes to mechanical and gas exchange abnormalities. However, radiographic and gas exchange abnormalities may not be sensitive *From the Departments of Pulmonary Medicine (Drs. Kunst, Noordegraaf, Postmus, and de Vries), Medical Physics and Informatics (Ms. Raaijmakers), and Intensive Care Medicine (Dr. Groeneveld), Institute for Cardiovascular Research, Academic Hospital “Vrije Universiteit,” Amsterdam; and Department of Intensive Care Medicine (Dr. Bakker), Hospital Centre Apeldoorn, Apeldoorn, The Netherlands. Supported by Glaxo Wellcome plc, The Netherlands. Manuscript received November 11, 1998; revision accepted June 10, 1999. Correspondence to: Peter W. A. Kunst, MD, Academic Hospital “Vrije Universiteit,” Department of Pulmonary Medicine, 1007 MB Amsterdam, The Netherlands
and specific and may not accurately reflect the extent of pulmonary edema.1 Assessing the amount of extravascular lung water (EVLW) at the bedside is useful because the amount of directly measured pulmonary edema may have prognostic significance in ARF, and amelioration of the edema may decrease morbidity and mortality.2– 4 On the other hand, assessing the amount of EVLW at the bedside is not simple. For this purpose, the thermal dye double indicator dilution technique (TDD) is used, but the invasiveness of the technique, which utilizes two catheters, has hampered widespread clinical application. Despite limitations inherent to the technique, including the potential underestimation of EVLW in CHEST / 116 / 6 / DECEMBER, 1999
1695
poorly perfused lung regions,5 the technique is considered as the “gold standard” for measuring EVLW.5,6 In fact, EVLW measured by TDD may largely reflect gravimetric EVLW measurements at autopsy of ARF patients.6 Noninvasive techniques used to assess the amount of EVLW include the transthoracic electrical impedance technique.7 However, although simple and noninvasive, results from studies that assess the amount of EVLW with this technique are variable.8 –11 The influence of blood volume8 and ventilation on the measured impedance change8 are major confounders in the assessment of the amount of EVLW by transthoracic electrical impedance. Furthermore, transthoracic electrical impedance does not allow one to differentiate between EVLW and pleural fluid. Therefore, it can be considered insufficiently sensitive and accurate for quantification of the amount of EVLW.8 Recently, a new noninvasive electrical impedance imaging technique, electrical impedance tomography (EIT), has been developed, which has the potential to overcome most of the limitations of transthoracic electrical impedance measurements. In contrast to transthoracic electrical impedance, EIT is an imaging technique that can visualize the electrical impedance changes caused by the ventilation in a two-dimensional transverse plane.12,13 The measured impedance changes due to ventilation have a linear relationship with tidal volume (Vt),12 and different regions can be analyzed quantitatively by selecting specific areas of interest in the measured plane.13,14 Since it is known that in ARDS, lung density increases from the ventral to the dorsal lung regions15 in the supine position, we hypothesized that EIT might be able to detect EVLW in ARF when ventilation-induced impedance changes of the anterior and the posterior parts of the lung are compared. Therefore, EIT was compared with TDD in patients with noncardiogenic ARF.
Materials and Methods Patients The study was performed at the ICU of the Saint Lucas Hospital in Apeldoorn, The Netherlands. Fourteen consecutive patients with a clinical diagnosis of noncardiogenic ARF were included in the study. The diagnosis was based on the presence of respiratory distress associated with dyspnea and tachypnea, hypoxemia, and bilateral and diffuse opacities on the chest roentgenogram in the absence of an elevated pulmonary capillary wedge pressure. Patients received mechanical ventilation with pressure control ventilation (Servo 900C; Siemens-Elema AB; Solna, Sweden). The levels of positive end-expiratory pressure (PEEP) and fraction of inspired oxygen (Fio2) were chosen according to clinical requirements. According to the lung injury 1696
score (LIS), patients were divided into two groups. The LIS was computed as described by Murray and coworkers16 using the number of quadrants involved on the chest roentgenogram, the Pao2/Fio2 ratio, the level of PEEP, and the quasi-static compliance (measured as Vt divided by inspiratory plateau pressure minus total end-expiratory pressure); each scored from zero to four. The mean of the scores gives the LIS. Patients with a score . 2.5 were defined as having ARDS.16 Patients with a LIS # 2.5 were defined as having acute lung injury (ALI). Patients with a history of chronic pulmonary disease were excluded because the impedance signal may be confounded by relatively poorly ventilated areas in emphysematous lungs. Patients were in a supine position. The protocol was approved by the local ethics committee, and informed consent was obtained from the closest relative of each patient. Protocol After inclusion in the study, a fiberoptic catheter was placed in the femoral artery for the detection of EVLW by TDD. Afterwards, 16 electrodes were attached at the third intercostal level, and two EIT measurements were performed. During the measurements, the PEEP level and the Vt were kept constant. Within 10 min after the EIT measurements, one EVLW measurement with TDD was performed. All of the described measurements were repeated daily until any one of the following occurred: the patient died, EVLW became , 10 mL/kg, or the fiberoptic catheter could not receive an accurate signal anymore due to damage. EIT: In this study, EIT measurements were performed using a portable data acquisition system (Sheffield Applied Potential Tomograph, DAS-01P, Mark I; IBEES; Sheffield, UK), which has been described before.17,18 In the present study, the 16 electrodes (Meditrace; Technomed; Beek, The Netherlands) were equidistantly spaced around the thorax. Impedance measurements were performed at the third intercostal level, with the patient in a supine position. A source that generates an alternating current (50 KHz, 5 mA peak to peak) was used to measure impedance. Data sets, consisting of 50 images, with an interval time of 1.13 s were recorded. One image consisted of 10 averaged data collection cycles used to minimize artifacts in impedance. Each image visualizes the impedance distribution within the thorax. Because inspiration and expiration of air cause relatively large impedance changes, these changes can be imaged. In the sequence of images, the impedance changes during ventilation can be visualized and studied. By defining a region of interest (ROI), specific areas can be examined. The computer calculated impedance changes within the ROI over the whole sequence of images. The average impedance change in this ROI is plotted in a curve as a function of time in order to show the impedance changes during ventilation. Because the reconstruction algorithm of the EIT is based on normalized differences, the average impedance change has no unit and is expressed as an arbitrary unit. Image Analysis: As expected, ventilation-induced impedance changes increased after the application of PEEP.19 To avoid these PEEP-associated changes, we chose the region where only high ventilation-induced impedance changes (the center of the image: 128 pixels) are seen as ROI, thereby assuming that alveolar recruitment in this area by PEEP would be minimal. In ARDS, lung density increases from the ventral to the dorsal lung regions15 in the supine position and an increased amount of EVLW causes compression atelectasis in the posterior part.20 Because EIT can visualize regional ventilation,13,14 we hypothesized that differences in the ventilation-induced impedance changes occurring between the anterior part and posterior part of the lungs may provide information about the EVLW content. Clinical Investigations in Critical Care
Therefore, the ROI was divided in an anterior half and a posterior half (Fig 1). We analyzed the curves of the whole ROI and those of the anterior half and the posterior half separately. From the obtained curves, the maximal difference in impedance between inspiration and expiration was calculated, and it was called the ventilation-induced impedance change. To obtain information about the difference in ventilation between the anterior and posterior parts of the lung, the ventilation-induced impedance change of the whole area of interest was divided by the ventilation-induced impedance change of the anterior half. The measured ventilation-induced impedance change of the posterior part of the lung may be difficult to assess and might be miscalculated due to the compression atelectasis occurring in the posterior part of the lung when edema increases.20 By taking the whole area of interest, information about the posterior part is included; therefore, theoretically, information about the difference in ventilation in the posterior and anterior parts of the lung following edema accumulation should be obtained. Thus, IR 5 ICw/ICa 5 (0.5[ICa 1 ICp])/ICa, where IR is the impedance ratio; ICw is the ventilation-induced impedance change between inspiration and expiration measured at the whole ROI; ICa is the ventilationinduced impedance change measured at the selected anterior part of the lung; and ICp is the ventilation-induced impedance change measured at the selected posterior part of the lung. TDD: In this study, TDD assessment of the amount of EVLW was performed using the COLD System (Pulsion Medical Systems; Munich, Germany). The principle is based on the injection of two indicators (indocyanine green [ICG] and cold glucose 5% [4°C]), with detection of their dilution curves after passage through the pulmonary circulation.5,21,22 The ICG and glucose (10 to 12 mL; 1 mg ICG was dissolved in 1 mL glucose 5%) are administered in the right atrium and detected in the femoral artery. ICG is protein-bound and therefore confined to the intravascular space. The thermal indicator distributes in the extravascular compartment. A specially designed thermistortipped fiberoptic catheter (Pulsion type) was used to detect both indicators in the femoral artery. In this way, the dye concentration is measured in vivo. The difference in area under the indicator dilution curves yields the volume outside the pulmonary capillaries, which is the volume of the EVLW.5,21,22 The EVLW is expressed in mL/kg body weight. Normal values range from 5 to 10 mL/kg. Patients are expected to have mild edema when the
values range from 10 to 20 mL/kg, and they are expected to have severe pulmonary edema when values are . 20 mL/kg.5 Statistical Analysis Data are presented as mean 6 SD. In order to depict the relationship between the IR and the values of EVLW obtained with TDD, linear regression analysis was used. The mean and SD of the differences between two consecutive EIT measurements were calculated and plotted against the mean values in order to visualize reproducibility. Furthermore, the coefficient of variation (CV) and the reliability coefficient (RC) between both measurements were calculated: CV is the SD of the difference within the repeated measurements divided by the mean of both measurements, whereas the RC is the variance of repeated measurements divided by the variance between repeated measurements and the variance of the difference within the repeated measurements. Two independent observers blinded for the EVLW measurements by the TDD analyzed the EIT measurements. Interobserver variation was studied by calculating Pearson’s correlation coefficient. Receiver operator characteristic (ROC) analysis was used to assess the optimal cutoff level of the IR to diagnose an increased amount of EVLW (. 10 mL/kg). The ROC curve shows the calculated sensitivity and specificity for a test over a range of cutoff points and can be used to determine the best cutoff point. The closer the area under the curve to 1, the better the diagnostic performance of a test. To investigate whether the IR has a relation with the severity of the disease, Pearson’s correlation analysis was used to investigate the relation between the LIS and the EVLW content and the IR. To underline the assumption that the IR is independent of the level of PEEP, the relation between PEEP and the EVLW content assessed by TDD and the IR was investigated by Pearson’s correlation analysis. Also, in three patients, two levels of PEEP were applied to a constant amount of EVLW. First, EIT measurements were performed at the level of PEEP necessary according to clinical conditions. Then, EIT measurements were performed after 10 min at zero PEEP. Because the variability of PEEP and LIS in the daily measurements made on the same subject will probably be small, only the day 1 measurements were used in both analyses in order to avoid any clusters of points.23 A p value , 0.05 was considered as statistically significant.
Results Patients
Figure 1. An image obtained by EIT at maximal inspiration in a patient with ARDS. The letters R, L, A, and P represent the right side, left side, anterior side, and posterior side of the image, respectively. The area of interest (indicated with the white lines) is divided into a posterior (post) part and an anterior (ant) part of the lungs. Next to the image, the curve, which shows the impedance changes occurring in the area of interest during inspiration and expiration, is depicted.
Eight patients were classified as having ARDS (LIS . 2.5), whereas six patients had ALI (LIS # 2.5). All patients had ARF resulting from sepsis: six after an abdominal operation, two due to a trauma, five following a severe pneumococcal pneumonia, and one following a urinary tract infection. The general characteristics of the patients are reported in Table 1. Two of the 14 patients (14%) died during their stay in the ICU. Both deaths occurred in the ARDS group (mortality, 25%). One patient (patient 7) died within 3 days after the onset of ARF. Another patient died within 6 days after the onset of ARF (patient 8). The remaining 12 patients could be extubated and discharged from the ICU and were considered as survivors. CHEST / 116 / 6 / DECEMBER, 1999
1697
Table 1—Characteristics of the Patients* Patient
Age, yr
Sex
Etiology of ARF
Pao2/Fio2, mm Hg
Compliance, cm H2O/mL
PEEP, cm H2O
X-ray
LIS
PCWP, mm Hg
Survival
1 2 3 4 5 6 7 8 9 10 11 12 13 14
59 68 75 39 74 76 57 42 68 26 41 59 30 36
Female Male Male Male Male Female Male Female Male Male Male Male Male Male
Sepsis pneumonia Abdominal sepsis Sepsis pneumonia Trauma Trauma Sepsis pneumonia Abdominal sepsis Sepsis pneumonia Abdominal sepsis Abdominal sepsis Abdominal sepsis Abdominal sepsis Urinary sepsis Sepsis pneumonia
63 152 316 254 103 115 162 133 238 165 95 71 106 283
16 48 45 60 32 58 28 12 50 35 22 17 29 40
6 5 5 5 12 8 12 12 5 10 12 8 11 10
2 2 3 2 3 2 4 2 4 3 4 2 3 3
2.75 1.75 1.25 1 3 2 3.25 3 1.75 2.75 3.5 2.75 2.75 2
14 8 12 5 11 18 4 9 11 14 10 12 8 12
yes yes yes yes yes yes no no yes yes yes yes yes yes
*Pao2/Fio2 5 hypoxemia score; PCWP 5 pulmonary capillary wedge pressure; X-ray 5 chest roentgenogram (results are the number of quadrants in which alveolar consolidation occurs).
Assessment of EVLW Twenty-nine measurements in 14 patients could be performed. In two patients, no second day follow-up measurement could be performed due to damage to the fiberoptic catheter (patients 1 and 6); in one patient, it was impossible to collect correct EIT data due to the existence of severe subcutaneous edema (patient 7). Nine patients were monitored during 2 days; seven of them were monitored because EVLW was , 10 mL/kg, and two of them were monitored because of damage to the fiberoptic catheter. Patients 8 and 12 were measured during 5 and 3 days, respectively: patient 8 died, and the EVLW content in patient 12 reached 10 mL/kg. At admission, the EVLW assessed by TDD was 11.2 6 7.0 mL/kg, and the IR was 0.68 6 0.15. As shown in Figure 2, bottom right, D, a significant correlation was found between the IR and the EVLW measured with TDD (r 5 0.85; p , 0.001). To test the hypothesis that EIT is useful for assessing the amount of EVLW, the relation between the ventilation-induced impedance changes of the anterior part and those of the posterior part with the TDD EVLW was examined separately. The correlation between the ventilation-induced impedance change measured at the anterior part and EVLW was poor (r 5 2 0.40; p 5 0.04; Fig 2, top right, B), whereas no correlation between the ventilation-induced impedance changes measured over both lungs (Fig 2, top left, A; r 5 0.11; p 5 0.57) or at the posterior part and EVLW existed (Fig 2, bottom left, C; r 5 2 0.27; p 5 0.27). A significant correlation also existed between the changes in EVLW measured by TDD and EIT (r 5 0.85; p , 0.005). Figure 3 shows follow-up measurements at 2 1698
consecutive days in 11 patients for TDD and EIT. In 3 of the 11 patients in whom repeated measurements of EVLW were made, TDD showed changes that were not seen with EIT (all in the normal range). Figure 4 depicts the curves of EIT and TDD for the monitoring of the EVLW content in patient 8 during 5 consecutive days. A similar course of the IR and TDD is shown. Figure 5 depicts a Bland-Altman plot to show the differences between the two repeated measurements of EIT against their means. Two reproducible measurements could not be obtained in two patients because of the disturbing influences of subcutaneous emphysema and edema on EIT measurements, and these were excluded from the analysis. The calculated CV was 4.1%, whereas the RC was 92%. A significant correlation of 0.91 (p , 0.001) between the analyses by both observers was found. To assess the diagnostic performance of EIT in the ROC curve, an EVLW of 10 mL/kg was regarded as the reference because this value is regarded as the upper limit of normal.5 The obtained ROC curve (Fig 6) shows an area under the curve of 0.86 (p , 0.005), which indicates a good discriminating power of EIT for the assessment of EVLW. The sensitivity, specificity, and positive and negative predictive values were calculated for several cutoff points. At an IR of 0.60, a sensitivity for a supranormal EVLW of 100% was reached. The accompanying specificity and positive and negative predicted values are 53%, 67%, and 100%, respectively. When an IR of 0.64 was chosen as the cutoff level, the sensitivity was 93%, the specificity was 87%, and the positive and negative predicted values were 87% and 93%, respectively. Clinical Investigations in Critical Care
Figure 2. Scatter plots showing the relation between EIT and thermal dye dilution in the assessment of the amount of EVLW in both lungs (top left, A), the anterior part of the lungs (top right, B), and the posterior part of the lungs (bottom left, C); and the assessment of an IR in which both components are incorporated (bottom right, D).
Respiratory Correlates No significant correlation was found between the LIS and EVLW obtained by TDD. Also, the amount of EVLW was not significantly greater in the ARDS group (n 5 8) in comparison with the ALI group (n 5 6; 12.9 6 6.5 vs 8.7 6 7.5, respectively; p 5 0.18). The IR showed no significant correlation with LIS, and also, there was no significant difference in the IR between the ARDS group and the ALI group (0.70 6 0.15 vs 0.64 6 0.17, respectively; p 5 0.41). No significant correlation existed between PEEP and the EVLW obtained with TDD or the IR (r 5 0.27, p 5 0.35 and r 5 0.20, p 5 0.49, respectively). In the three patients in whom the influence of PEEP on the IR was examined by performing EIT measurements on the initial PEEP level and at zero PEEP, the difference in IRs at the different PEEP levels did not exceed 10% (Table 2).
Discussion Our data show that EIT, by using an IR that is based on ventilation-induced impedance changes in the anterior and the posterior parts of the lung, is a noninvasive technique that may be a reasonable estimate of EVLW in ARF. The explanation for the increase in the IR, and thus the presence of EVLW, which reflects the ventilation-induced impedance changes of the posterior part divided by those of the anterior part of the lungs, might be twofold. First, ALI is known to result in a decreased gas/tissue ratio in the anterior and the posterior parts of the lung, as compared to the healthy state.24 Therefore, a substantial decrease in ventilation-induced impedance changes can be expected in the anterior part. And second, as lung weight increases with an increasing EVLW, the lung CHEST / 116 / 6 / DECEMBER, 1999
1699
Figure 3. Follow-up measurements of the first 2 consecutive days of the amount of EVLW obtained with thermal dye dilution and EIT. Each symbol reflects a patient.
is moved to the back in the supine position due to gravitational forces.20 This is reflected by a bigger decrease in the ventilation-induced impedance change in the anterior part of the lung than that of the posterior part because the effect of gravity on the dense lung results in a greater than normal gradient from the top to the bottom of the lung (Fig 2, top right, B and bottom left, C).15,24 Thus, in the EIT image, the lung is moved to the back, and because the EIT image was divided into an anterior and posterior part of the lung by a standardized ROI analysis, the ventilation-induced impedance changes in the anterior part of the lung may decrease and those in the posterior part of the lung may even
Figure 4. The follow-up measurements during 5 days of patient 8 for thermal dye dilution and EIT. 1700
increase due to a shift in the amount of alveoli in the parts of the lung selected, thereby increasing the IR. Theoretically, several factors might influence the IR. First, PEEP may increase the lung volume predominantly in the dependent parts of the lung19,25 and thereby increase the ventilationinduced impedance changes in those parts of the lung. In order to avoid this interference, we chose the center of the image in which the highest ventilation-induced impedance changes occur as the ROI for our analysis. The lack of significant correlation between PEEP and the IR, and the fact that in three patients, despite the change in PEEP (although small) and an unchanged EVLW, the IR remained within a 10% change, prove the independency of the IR in relation to the PEEP. Second, pleural effusion interfered with transthoracic impedance measurements of EVLW. Because the presence of EVLW by EIT is determined by the gas inflation differences, pleural effusion will not interfere. Third, because EIT uses an electrical current, circumstances that influence conductivities of tissue might disturb the assessment of accurate EIT-data (ie, subcutaneous edema and subcutaneous emphysema). In this study, nine patients suffered from subcutaneous edema, but the latter prevented meaningful data collection in only one patient. Finally, because the IR is based on gas inflation differences, diseases that alter the distribution of ventilation (emphysema, lung cancer, or pneumonectomy) might disturb the interpretation of the IR as an indication of the presence of EVLW. No patient in our study was known to suffer from any of these coincidental diseases; as a result, their effect remains unknown. Clinical Investigations in Critical Care
Figure 5. Differences of repeated measurements vs the mean of both measurements show the reproducibility of EIT. Almost all measurements are within 23SDs (dashed lines). The calculated CV and RC is 4.1% and 92%, respectively.
In our study, no correlation between the EVLW content, as determined by TDD or the IR, and the LIS was found. Based on this result, EIT and TDD may not be used to judge the severity of ARF, although the literature shows that the EVLW content can serve as a tool to judge the severity of ARF.2,26 Also, the clinical benefit of measuring EVLW in patients with ARF is controversial. Whereas some authors have showed that pulmonary edema may have prognostic importance3,4 and that reducing pulmonary edema may improve outcome,3,7 others have not been able to demonstrate an
association between EVLW and outcome in ARF.27,28 Despite these limitations, the use of EIT to assess the amount of EVLW might still be desirable for several reasons. First, EIT was compared to an invasive “gold standard.” Although we know that TDD has its limitations (ie, underestimation in patients with intravascular pulmonary shunts5,6), it is a technique that has a very good correlation with gravimetric techniques and is clinically available. In this clinical study, EIT showed good results compared to TDD and, therefore, is promising. A study in which EIT is compared to gravimetric techniques is still necessary. Second, because TDD is invasive and expensive, no good follow-up studies for clinical outcome are possible. EIT is inexpensive and noninvasive, thereby allowing for the opportunity to do more research that will end the controversy over EVLW and outcome in literature. In conclusion, the present study shows that EIT is a noninvasive technique that may be used to reasonably estimate the presence of EVLW in ARF if the IR based on ventilation-induced impedance changes between the posterior and anterior part of the lung is applied.
Table 2—Differences in IR at Different PEEP Levels*
Figure 6. An ROC curve showing the sensitivity and specificity in the diagnosis of an increased amount of EVLW for different IRs. The area under the curve is 0.86 (p , 0.005).
Patient
EVLW, mL/kg
PEEP, cm H2O
IR
1
NA
2
4.8
3
6.5
6 0 5 0 5 0
0.58 0.56 0.48 0.50 0.59 0.63
% Change in IR 3.4 4.2 6.7
*NA 5 not available, due to fiberoptic damage of the TDD catheter. CHEST / 116 / 6 / DECEMBER, 1999
1701
References 1 Miniati M, Pistolesi M, Milne E, et al. Detection of lung edema. Crit Care Med 1987; 15:1146 –1155 2 Demling RH, Lalonde C, Ikegami K. Pulmonary edema: pathophysiology, methods of measurement, and clinical importance in acute respiratory failure. New Horiz 1993; 1:371– 380 3 Schuller D, Mitchell JP, Calandrino FS, et al. Fluid balance during pulmonary edema: is fluid gain a marker or a cause of poor outcome? Chest 1991; 100:1068 –1075 4 Eisenberg PR, Hansborough JR, Anderson D, et al. A prospective study of lung water measurements during patient management in an intensive care unit. Am Rev Respir Dis 1987; 136:662– 668 5 Bo¨ck J, Lewis FR. Clinical relevance of lung water measurement with the thermal dye dilution technique. J Surg Res 1990; 48:254 –265 6 Mihm FG, Feeley TW, Jamieson SW. Thermal dye double indicator dilution measurement of lung water in man: comparison with gravimetric measurements. Thorax 1987; 42: 72–76 7 Staub N. Clinical use of lung water measurements: report of a workshop. Chest 1986; 90:589 –594 8 Fein A, Grossman RF, Jones JG, et al. Evaluation of transthoracic electrical impedance in the diagnosis of pulmonary edema. Circulation 1979; 60:1156 –1160 9 Spinale FG, Reines HD, Cook MC, et al. Noninvasive estimation of extravascular lung water using bioimpedance. J Surg Res 1989; 47:535–540 10 Korsten HHM, Leusink JA, Spierdijk J, et al. Pulmonary shunting after cardiopulmonary bypass. Eur Heart J 1989; 10(suppl):1–5 11 Jonsson F, Madsen P, Jorgensen LG, et al. Thoracic electrical impedance and fluid balance during aortic surgery. Acta Anaesthesiol Scand 1995; 39:513–517 12 Harris ND, Suggett AJ, Barber DC, et al. Applications of applied potential tomography (APT) in respiratory medicine. Clin Phys Physiol Meas 1987; 8(suppl):155–165 13 Hahn G, Sipinkova I, Baisch F, et al. Changes in thoracic impedance distribution under different ventilatory conditions. Physiol Meas 1995; 16:A161–A173 14 Kunst PW, Vonk Noordegraaf A, Hoekstra O, et al. Ventilation and perfusion imaging by electrical impedance tomography: a comparison with radionuclide scanning. Physiol Meas 1998; 19:481– 490
1702
15 Gattinoni L, Pelosi P, Vitale G, et al. Body position changes redistribute lung computed tomographic density in patients with acute respiratory failure. Anesthesiology 1991; 74:15–23 16 Murray JF, Matthay MA, Luce JM, et al. An expanded definition of the adult respiratory distress syndrome. Am Rev Respir Dis 1988; 138:720 –723 17 Brown BH, Barber DC. Possibilities and problems of realtime imaging of tissue resistivity. Clin Phys Physiol Meas 1988; 9(suppl):121–125 18 Smith RWM, Freeston IL, Brown BH. A real time electrical impedance tomography system for clinical use: design and preliminary results. IEEE Trans Biomed Eng 1995; 40:133– 139 19 Kunst PW, de Vries PMJM, Postmus PE, et al. Evaluation of electrical impedance tomography in the measurement of PEEP-induced changes in lung volume. Chest 1999; 115: 1102–1106 20 Pelosi P, Crotti S, Brazzi L, et al. Computed tomography in adult respiratory distress syndrome: what has it taught us? Eur Respir J 1996; 9:1055–1062 21 McLuckie A. The COLD system of hemodynamic monitoring. Int Care World 1996; 13:24 –28 22 Sivak ED, Wiedemann HP. Clinical measurement of extravascular lung water. Crit Care Clin 1986; 2:511–526 23 Bland JM, Altman DG. Correlation, regression, and repeated data. BMJ 1994; 308:896 24 Gattinoni L, Bombino M, Pelosi P, et al. Lung structure and function in different stages of severe adult respiratory distress syndrome. JAMA 1994; 271:1772–1779 25 Gattinoni L, Pelosi P, Crotti S, et al. Effects of positive end-expiratory pressure on regional distribution of tidal volume and recruitment in adult respiratory distress syndrome. Am J Respir Crit Care Med 1995; 151:1807–1814 26 Bigatello LM, Zapol WM. New approaches to acute lung injury. Br J Anaesth 1996; 77:99 –109 27 Mitchell JP, Schuller D, Calandrino FS, et al. Improved outcome based on fluid management in critically ill patients requiring pulmonary artery catheterization. Am Rev Respir Dis 1992; 145:990 –998 28 Brigham KL, Kariman K, Harris TR, et al. Correlation of oxygenation with vascular permeability-surface area but not with lung water in humans with acute respiratory failure and pulmonary edema. J Clin Invest 1983; 72:339 –349
Clinical Investigations in Critical Care
Study objectives: To establish the value of electrical impedance tomography (EIT) in assessing pulmonary edema in noncardiogenic acute respiratory failure (ARF), as compared to the thermal dye double indicator dilution technique (TDD). Design: Prospective clinical study. Setting: ICU of a general hospital. Patients: Fourteen ARF patients. Interventions: In order to use the TDD to determine the amount of extravascular lung water (EVLW), a fiberoptic catheter was placed in the femoral artery. Measurements and main results: Fourteen consecutive ARF patients receiving mechanical ventilation were measured by EIT and TDD. EIT visualizes the impedance changes caused by the ventilation in two-dimensional image planes. An impedance ratio (IR) of the ventilation-induced impedance changes of a posterior and an anterior part of the lungs was used to indicate the amount of EVLW. For the 29 measurements in 14 patients, a significant correlation between EIT and TDD (r 5 0.85; p < 0.001) was found. The EIT reproducibility was good. The diagnostic value of the method was tested by receiver operator characteristic analysis, with 10 mL/kg of EVLW considered as the upper limit of normal. At a cutoff level of the IR of 0.64, the IR had a sensitivity of 93%, a specificity of 87%, and a positive predictive value of 87% for a supranormal amount of EVLW. Follow-up measurements were performed in 11 patients. A significant correlation was found between the changes in EVLW measured with EIT and TDD (r 5 0.85; p < 0.005). Conclusion: We conclude that EIT is a noninvasive technique for reasonably estimating the amount of EVLW in noncardiogenic ARF. (CHEST 1999; 116:1695–1702) Key words: ARDS; electrical impedance; extravascular lung water; thermal dye dilution; tomography Abbreviations: ALI 5 acute lung injury; ARF 5 acute respiratory failure; CV 5 coefficient of variation; EIT 5 electrical impedance tomography; EVLW 5 extravascular lung water; Fio2 5 fraction of inspired oxygen; ICG 5 indocyanine green; IR 5 impedance ratio; LIS 5 lung injury score; PEEP 5 positive end-expiratory pressure; RC 5 reliability coefficient; ROC 5 receiver operator characteristic; ROI 5 region of interest; TDD 5 thermal dye double indicator dilution technique; Vt 5 tidal volume
many cases of acute respiratory failure (ARF), I npulmonary edema contributes to mechanical and gas exchange abnormalities. However, radiographic and gas exchange abnormalities may not be sensitive *From the Departments of Pulmonary Medicine (Drs. Kunst, Noordegraaf, Postmus, and de Vries), Medical Physics and Informatics (Ms. Raaijmakers), and Intensive Care Medicine (Dr. Groeneveld), Institute for Cardiovascular Research, Academic Hospital “Vrije Universiteit,” Amsterdam; and Department of Intensive Care Medicine (Dr. Bakker), Hospital Centre Apeldoorn, Apeldoorn, The Netherlands. Supported by Glaxo Wellcome plc, The Netherlands. Manuscript received November 11, 1998; revision accepted June 10, 1999. Correspondence to: Peter W. A. Kunst, MD, Academic Hospital “Vrije Universiteit,” Department of Pulmonary Medicine, 1007 MB Amsterdam, The Netherlands
and specific and may not accurately reflect the extent of pulmonary edema.1 Assessing the amount of extravascular lung water (EVLW) at the bedside is useful because the amount of directly measured pulmonary edema may have prognostic significance in ARF, and amelioration of the edema may decrease morbidity and mortality.2– 4 On the other hand, assessing the amount of EVLW at the bedside is not simple. For this purpose, the thermal dye double indicator dilution technique (TDD) is used, but the invasiveness of the technique, which utilizes two catheters, has hampered widespread clinical application. Despite limitations inherent to the technique, including the potential underestimation of EVLW in CHEST / 116 / 6 / DECEMBER, 1999
1695
poorly perfused lung regions,5 the technique is considered as the “gold standard” for measuring EVLW.5,6 In fact, EVLW measured by TDD may largely reflect gravimetric EVLW measurements at autopsy of ARF patients.6 Noninvasive techniques used to assess the amount of EVLW include the transthoracic electrical impedance technique.7 However, although simple and noninvasive, results from studies that assess the amount of EVLW with this technique are variable.8 –11 The influence of blood volume8 and ventilation on the measured impedance change8 are major confounders in the assessment of the amount of EVLW by transthoracic electrical impedance. Furthermore, transthoracic electrical impedance does not allow one to differentiate between EVLW and pleural fluid. Therefore, it can be considered insufficiently sensitive and accurate for quantification of the amount of EVLW.8 Recently, a new noninvasive electrical impedance imaging technique, electrical impedance tomography (EIT), has been developed, which has the potential to overcome most of the limitations of transthoracic electrical impedance measurements. In contrast to transthoracic electrical impedance, EIT is an imaging technique that can visualize the electrical impedance changes caused by the ventilation in a two-dimensional transverse plane.12,13 The measured impedance changes due to ventilation have a linear relationship with tidal volume (Vt),12 and different regions can be analyzed quantitatively by selecting specific areas of interest in the measured plane.13,14 Since it is known that in ARDS, lung density increases from the ventral to the dorsal lung regions15 in the supine position, we hypothesized that EIT might be able to detect EVLW in ARF when ventilation-induced impedance changes of the anterior and the posterior parts of the lung are compared. Therefore, EIT was compared with TDD in patients with noncardiogenic ARF.
Materials and Methods Patients The study was performed at the ICU of the Saint Lucas Hospital in Apeldoorn, The Netherlands. Fourteen consecutive patients with a clinical diagnosis of noncardiogenic ARF were included in the study. The diagnosis was based on the presence of respiratory distress associated with dyspnea and tachypnea, hypoxemia, and bilateral and diffuse opacities on the chest roentgenogram in the absence of an elevated pulmonary capillary wedge pressure. Patients received mechanical ventilation with pressure control ventilation (Servo 900C; Siemens-Elema AB; Solna, Sweden). The levels of positive end-expiratory pressure (PEEP) and fraction of inspired oxygen (Fio2) were chosen according to clinical requirements. According to the lung injury 1696
score (LIS), patients were divided into two groups. The LIS was computed as described by Murray and coworkers16 using the number of quadrants involved on the chest roentgenogram, the Pao2/Fio2 ratio, the level of PEEP, and the quasi-static compliance (measured as Vt divided by inspiratory plateau pressure minus total end-expiratory pressure); each scored from zero to four. The mean of the scores gives the LIS. Patients with a score . 2.5 were defined as having ARDS.16 Patients with a LIS # 2.5 were defined as having acute lung injury (ALI). Patients with a history of chronic pulmonary disease were excluded because the impedance signal may be confounded by relatively poorly ventilated areas in emphysematous lungs. Patients were in a supine position. The protocol was approved by the local ethics committee, and informed consent was obtained from the closest relative of each patient. Protocol After inclusion in the study, a fiberoptic catheter was placed in the femoral artery for the detection of EVLW by TDD. Afterwards, 16 electrodes were attached at the third intercostal level, and two EIT measurements were performed. During the measurements, the PEEP level and the Vt were kept constant. Within 10 min after the EIT measurements, one EVLW measurement with TDD was performed. All of the described measurements were repeated daily until any one of the following occurred: the patient died, EVLW became , 10 mL/kg, or the fiberoptic catheter could not receive an accurate signal anymore due to damage. EIT: In this study, EIT measurements were performed using a portable data acquisition system (Sheffield Applied Potential Tomograph, DAS-01P, Mark I; IBEES; Sheffield, UK), which has been described before.17,18 In the present study, the 16 electrodes (Meditrace; Technomed; Beek, The Netherlands) were equidistantly spaced around the thorax. Impedance measurements were performed at the third intercostal level, with the patient in a supine position. A source that generates an alternating current (50 KHz, 5 mA peak to peak) was used to measure impedance. Data sets, consisting of 50 images, with an interval time of 1.13 s were recorded. One image consisted of 10 averaged data collection cycles used to minimize artifacts in impedance. Each image visualizes the impedance distribution within the thorax. Because inspiration and expiration of air cause relatively large impedance changes, these changes can be imaged. In the sequence of images, the impedance changes during ventilation can be visualized and studied. By defining a region of interest (ROI), specific areas can be examined. The computer calculated impedance changes within the ROI over the whole sequence of images. The average impedance change in this ROI is plotted in a curve as a function of time in order to show the impedance changes during ventilation. Because the reconstruction algorithm of the EIT is based on normalized differences, the average impedance change has no unit and is expressed as an arbitrary unit. Image Analysis: As expected, ventilation-induced impedance changes increased after the application of PEEP.19 To avoid these PEEP-associated changes, we chose the region where only high ventilation-induced impedance changes (the center of the image: 128 pixels) are seen as ROI, thereby assuming that alveolar recruitment in this area by PEEP would be minimal. In ARDS, lung density increases from the ventral to the dorsal lung regions15 in the supine position and an increased amount of EVLW causes compression atelectasis in the posterior part.20 Because EIT can visualize regional ventilation,13,14 we hypothesized that differences in the ventilation-induced impedance changes occurring between the anterior part and posterior part of the lungs may provide information about the EVLW content. Clinical Investigations in Critical Care
Therefore, the ROI was divided in an anterior half and a posterior half (Fig 1). We analyzed the curves of the whole ROI and those of the anterior half and the posterior half separately. From the obtained curves, the maximal difference in impedance between inspiration and expiration was calculated, and it was called the ventilation-induced impedance change. To obtain information about the difference in ventilation between the anterior and posterior parts of the lung, the ventilation-induced impedance change of the whole area of interest was divided by the ventilation-induced impedance change of the anterior half. The measured ventilation-induced impedance change of the posterior part of the lung may be difficult to assess and might be miscalculated due to the compression atelectasis occurring in the posterior part of the lung when edema increases.20 By taking the whole area of interest, information about the posterior part is included; therefore, theoretically, information about the difference in ventilation in the posterior and anterior parts of the lung following edema accumulation should be obtained. Thus, IR 5 ICw/ICa 5 (0.5[ICa 1 ICp])/ICa, where IR is the impedance ratio; ICw is the ventilation-induced impedance change between inspiration and expiration measured at the whole ROI; ICa is the ventilationinduced impedance change measured at the selected anterior part of the lung; and ICp is the ventilation-induced impedance change measured at the selected posterior part of the lung. TDD: In this study, TDD assessment of the amount of EVLW was performed using the COLD System (Pulsion Medical Systems; Munich, Germany). The principle is based on the injection of two indicators (indocyanine green [ICG] and cold glucose 5% [4°C]), with detection of their dilution curves after passage through the pulmonary circulation.5,21,22 The ICG and glucose (10 to 12 mL; 1 mg ICG was dissolved in 1 mL glucose 5%) are administered in the right atrium and detected in the femoral artery. ICG is protein-bound and therefore confined to the intravascular space. The thermal indicator distributes in the extravascular compartment. A specially designed thermistortipped fiberoptic catheter (Pulsion type) was used to detect both indicators in the femoral artery. In this way, the dye concentration is measured in vivo. The difference in area under the indicator dilution curves yields the volume outside the pulmonary capillaries, which is the volume of the EVLW.5,21,22 The EVLW is expressed in mL/kg body weight. Normal values range from 5 to 10 mL/kg. Patients are expected to have mild edema when the
values range from 10 to 20 mL/kg, and they are expected to have severe pulmonary edema when values are . 20 mL/kg.5 Statistical Analysis Data are presented as mean 6 SD. In order to depict the relationship between the IR and the values of EVLW obtained with TDD, linear regression analysis was used. The mean and SD of the differences between two consecutive EIT measurements were calculated and plotted against the mean values in order to visualize reproducibility. Furthermore, the coefficient of variation (CV) and the reliability coefficient (RC) between both measurements were calculated: CV is the SD of the difference within the repeated measurements divided by the mean of both measurements, whereas the RC is the variance of repeated measurements divided by the variance between repeated measurements and the variance of the difference within the repeated measurements. Two independent observers blinded for the EVLW measurements by the TDD analyzed the EIT measurements. Interobserver variation was studied by calculating Pearson’s correlation coefficient. Receiver operator characteristic (ROC) analysis was used to assess the optimal cutoff level of the IR to diagnose an increased amount of EVLW (. 10 mL/kg). The ROC curve shows the calculated sensitivity and specificity for a test over a range of cutoff points and can be used to determine the best cutoff point. The closer the area under the curve to 1, the better the diagnostic performance of a test. To investigate whether the IR has a relation with the severity of the disease, Pearson’s correlation analysis was used to investigate the relation between the LIS and the EVLW content and the IR. To underline the assumption that the IR is independent of the level of PEEP, the relation between PEEP and the EVLW content assessed by TDD and the IR was investigated by Pearson’s correlation analysis. Also, in three patients, two levels of PEEP were applied to a constant amount of EVLW. First, EIT measurements were performed at the level of PEEP necessary according to clinical conditions. Then, EIT measurements were performed after 10 min at zero PEEP. Because the variability of PEEP and LIS in the daily measurements made on the same subject will probably be small, only the day 1 measurements were used in both analyses in order to avoid any clusters of points.23 A p value , 0.05 was considered as statistically significant.
Results Patients
Figure 1. An image obtained by EIT at maximal inspiration in a patient with ARDS. The letters R, L, A, and P represent the right side, left side, anterior side, and posterior side of the image, respectively. The area of interest (indicated with the white lines) is divided into a posterior (post) part and an anterior (ant) part of the lungs. Next to the image, the curve, which shows the impedance changes occurring in the area of interest during inspiration and expiration, is depicted.
Eight patients were classified as having ARDS (LIS . 2.5), whereas six patients had ALI (LIS # 2.5). All patients had ARF resulting from sepsis: six after an abdominal operation, two due to a trauma, five following a severe pneumococcal pneumonia, and one following a urinary tract infection. The general characteristics of the patients are reported in Table 1. Two of the 14 patients (14%) died during their stay in the ICU. Both deaths occurred in the ARDS group (mortality, 25%). One patient (patient 7) died within 3 days after the onset of ARF. Another patient died within 6 days after the onset of ARF (patient 8). The remaining 12 patients could be extubated and discharged from the ICU and were considered as survivors. CHEST / 116 / 6 / DECEMBER, 1999
1697
Table 1—Characteristics of the Patients* Patient
Age, yr
Sex
Etiology of ARF
Pao2/Fio2, mm Hg
Compliance, cm H2O/mL
PEEP, cm H2O
X-ray
LIS
PCWP, mm Hg
Survival
1 2 3 4 5 6 7 8 9 10 11 12 13 14
59 68 75 39 74 76 57 42 68 26 41 59 30 36
Female Male Male Male Male Female Male Female Male Male Male Male Male Male
Sepsis pneumonia Abdominal sepsis Sepsis pneumonia Trauma Trauma Sepsis pneumonia Abdominal sepsis Sepsis pneumonia Abdominal sepsis Abdominal sepsis Abdominal sepsis Abdominal sepsis Urinary sepsis Sepsis pneumonia
63 152 316 254 103 115 162 133 238 165 95 71 106 283
16 48 45 60 32 58 28 12 50 35 22 17 29 40
6 5 5 5 12 8 12 12 5 10 12 8 11 10
2 2 3 2 3 2 4 2 4 3 4 2 3 3
2.75 1.75 1.25 1 3 2 3.25 3 1.75 2.75 3.5 2.75 2.75 2
14 8 12 5 11 18 4 9 11 14 10 12 8 12
yes yes yes yes yes yes no no yes yes yes yes yes yes
*Pao2/Fio2 5 hypoxemia score; PCWP 5 pulmonary capillary wedge pressure; X-ray 5 chest roentgenogram (results are the number of quadrants in which alveolar consolidation occurs).
Assessment of EVLW Twenty-nine measurements in 14 patients could be performed. In two patients, no second day follow-up measurement could be performed due to damage to the fiberoptic catheter (patients 1 and 6); in one patient, it was impossible to collect correct EIT data due to the existence of severe subcutaneous edema (patient 7). Nine patients were monitored during 2 days; seven of them were monitored because EVLW was , 10 mL/kg, and two of them were monitored because of damage to the fiberoptic catheter. Patients 8 and 12 were measured during 5 and 3 days, respectively: patient 8 died, and the EVLW content in patient 12 reached 10 mL/kg. At admission, the EVLW assessed by TDD was 11.2 6 7.0 mL/kg, and the IR was 0.68 6 0.15. As shown in Figure 2, bottom right, D, a significant correlation was found between the IR and the EVLW measured with TDD (r 5 0.85; p , 0.001). To test the hypothesis that EIT is useful for assessing the amount of EVLW, the relation between the ventilation-induced impedance changes of the anterior part and those of the posterior part with the TDD EVLW was examined separately. The correlation between the ventilation-induced impedance change measured at the anterior part and EVLW was poor (r 5 2 0.40; p 5 0.04; Fig 2, top right, B), whereas no correlation between the ventilation-induced impedance changes measured over both lungs (Fig 2, top left, A; r 5 0.11; p 5 0.57) or at the posterior part and EVLW existed (Fig 2, bottom left, C; r 5 2 0.27; p 5 0.27). A significant correlation also existed between the changes in EVLW measured by TDD and EIT (r 5 0.85; p , 0.005). Figure 3 shows follow-up measurements at 2 1698
consecutive days in 11 patients for TDD and EIT. In 3 of the 11 patients in whom repeated measurements of EVLW were made, TDD showed changes that were not seen with EIT (all in the normal range). Figure 4 depicts the curves of EIT and TDD for the monitoring of the EVLW content in patient 8 during 5 consecutive days. A similar course of the IR and TDD is shown. Figure 5 depicts a Bland-Altman plot to show the differences between the two repeated measurements of EIT against their means. Two reproducible measurements could not be obtained in two patients because of the disturbing influences of subcutaneous emphysema and edema on EIT measurements, and these were excluded from the analysis. The calculated CV was 4.1%, whereas the RC was 92%. A significant correlation of 0.91 (p , 0.001) between the analyses by both observers was found. To assess the diagnostic performance of EIT in the ROC curve, an EVLW of 10 mL/kg was regarded as the reference because this value is regarded as the upper limit of normal.5 The obtained ROC curve (Fig 6) shows an area under the curve of 0.86 (p , 0.005), which indicates a good discriminating power of EIT for the assessment of EVLW. The sensitivity, specificity, and positive and negative predictive values were calculated for several cutoff points. At an IR of 0.60, a sensitivity for a supranormal EVLW of 100% was reached. The accompanying specificity and positive and negative predicted values are 53%, 67%, and 100%, respectively. When an IR of 0.64 was chosen as the cutoff level, the sensitivity was 93%, the specificity was 87%, and the positive and negative predicted values were 87% and 93%, respectively. Clinical Investigations in Critical Care
Figure 2. Scatter plots showing the relation between EIT and thermal dye dilution in the assessment of the amount of EVLW in both lungs (top left, A), the anterior part of the lungs (top right, B), and the posterior part of the lungs (bottom left, C); and the assessment of an IR in which both components are incorporated (bottom right, D).
Respiratory Correlates No significant correlation was found between the LIS and EVLW obtained by TDD. Also, the amount of EVLW was not significantly greater in the ARDS group (n 5 8) in comparison with the ALI group (n 5 6; 12.9 6 6.5 vs 8.7 6 7.5, respectively; p 5 0.18). The IR showed no significant correlation with LIS, and also, there was no significant difference in the IR between the ARDS group and the ALI group (0.70 6 0.15 vs 0.64 6 0.17, respectively; p 5 0.41). No significant correlation existed between PEEP and the EVLW obtained with TDD or the IR (r 5 0.27, p 5 0.35 and r 5 0.20, p 5 0.49, respectively). In the three patients in whom the influence of PEEP on the IR was examined by performing EIT measurements on the initial PEEP level and at zero PEEP, the difference in IRs at the different PEEP levels did not exceed 10% (Table 2).
Discussion Our data show that EIT, by using an IR that is based on ventilation-induced impedance changes in the anterior and the posterior parts of the lung, is a noninvasive technique that may be a reasonable estimate of EVLW in ARF. The explanation for the increase in the IR, and thus the presence of EVLW, which reflects the ventilation-induced impedance changes of the posterior part divided by those of the anterior part of the lungs, might be twofold. First, ALI is known to result in a decreased gas/tissue ratio in the anterior and the posterior parts of the lung, as compared to the healthy state.24 Therefore, a substantial decrease in ventilation-induced impedance changes can be expected in the anterior part. And second, as lung weight increases with an increasing EVLW, the lung CHEST / 116 / 6 / DECEMBER, 1999
1699
Figure 3. Follow-up measurements of the first 2 consecutive days of the amount of EVLW obtained with thermal dye dilution and EIT. Each symbol reflects a patient.
is moved to the back in the supine position due to gravitational forces.20 This is reflected by a bigger decrease in the ventilation-induced impedance change in the anterior part of the lung than that of the posterior part because the effect of gravity on the dense lung results in a greater than normal gradient from the top to the bottom of the lung (Fig 2, top right, B and bottom left, C).15,24 Thus, in the EIT image, the lung is moved to the back, and because the EIT image was divided into an anterior and posterior part of the lung by a standardized ROI analysis, the ventilation-induced impedance changes in the anterior part of the lung may decrease and those in the posterior part of the lung may even
Figure 4. The follow-up measurements during 5 days of patient 8 for thermal dye dilution and EIT. 1700
increase due to a shift in the amount of alveoli in the parts of the lung selected, thereby increasing the IR. Theoretically, several factors might influence the IR. First, PEEP may increase the lung volume predominantly in the dependent parts of the lung19,25 and thereby increase the ventilationinduced impedance changes in those parts of the lung. In order to avoid this interference, we chose the center of the image in which the highest ventilation-induced impedance changes occur as the ROI for our analysis. The lack of significant correlation between PEEP and the IR, and the fact that in three patients, despite the change in PEEP (although small) and an unchanged EVLW, the IR remained within a 10% change, prove the independency of the IR in relation to the PEEP. Second, pleural effusion interfered with transthoracic impedance measurements of EVLW. Because the presence of EVLW by EIT is determined by the gas inflation differences, pleural effusion will not interfere. Third, because EIT uses an electrical current, circumstances that influence conductivities of tissue might disturb the assessment of accurate EIT-data (ie, subcutaneous edema and subcutaneous emphysema). In this study, nine patients suffered from subcutaneous edema, but the latter prevented meaningful data collection in only one patient. Finally, because the IR is based on gas inflation differences, diseases that alter the distribution of ventilation (emphysema, lung cancer, or pneumonectomy) might disturb the interpretation of the IR as an indication of the presence of EVLW. No patient in our study was known to suffer from any of these coincidental diseases; as a result, their effect remains unknown. Clinical Investigations in Critical Care
Figure 5. Differences of repeated measurements vs the mean of both measurements show the reproducibility of EIT. Almost all measurements are within 23SDs (dashed lines). The calculated CV and RC is 4.1% and 92%, respectively.
In our study, no correlation between the EVLW content, as determined by TDD or the IR, and the LIS was found. Based on this result, EIT and TDD may not be used to judge the severity of ARF, although the literature shows that the EVLW content can serve as a tool to judge the severity of ARF.2,26 Also, the clinical benefit of measuring EVLW in patients with ARF is controversial. Whereas some authors have showed that pulmonary edema may have prognostic importance3,4 and that reducing pulmonary edema may improve outcome,3,7 others have not been able to demonstrate an
association between EVLW and outcome in ARF.27,28 Despite these limitations, the use of EIT to assess the amount of EVLW might still be desirable for several reasons. First, EIT was compared to an invasive “gold standard.” Although we know that TDD has its limitations (ie, underestimation in patients with intravascular pulmonary shunts5,6), it is a technique that has a very good correlation with gravimetric techniques and is clinically available. In this clinical study, EIT showed good results compared to TDD and, therefore, is promising. A study in which EIT is compared to gravimetric techniques is still necessary. Second, because TDD is invasive and expensive, no good follow-up studies for clinical outcome are possible. EIT is inexpensive and noninvasive, thereby allowing for the opportunity to do more research that will end the controversy over EVLW and outcome in literature. In conclusion, the present study shows that EIT is a noninvasive technique that may be used to reasonably estimate the presence of EVLW in ARF if the IR based on ventilation-induced impedance changes between the posterior and anterior part of the lung is applied.
Table 2—Differences in IR at Different PEEP Levels*
Figure 6. An ROC curve showing the sensitivity and specificity in the diagnosis of an increased amount of EVLW for different IRs. The area under the curve is 0.86 (p , 0.005).
Patient
EVLW, mL/kg
PEEP, cm H2O
IR
1
NA
2
4.8
3
6.5
6 0 5 0 5 0
0.58 0.56 0.48 0.50 0.59 0.63
% Change in IR 3.4 4.2 6.7
*NA 5 not available, due to fiberoptic damage of the TDD catheter. CHEST / 116 / 6 / DECEMBER, 1999
1701
References 1 Miniati M, Pistolesi M, Milne E, et al. Detection of lung edema. Crit Care Med 1987; 15:1146 –1155 2 Demling RH, Lalonde C, Ikegami K. Pulmonary edema: pathophysiology, methods of measurement, and clinical importance in acute respiratory failure. New Horiz 1993; 1:371– 380 3 Schuller D, Mitchell JP, Calandrino FS, et al. Fluid balance during pulmonary edema: is fluid gain a marker or a cause of poor outcome? Chest 1991; 100:1068 –1075 4 Eisenberg PR, Hansborough JR, Anderson D, et al. A prospective study of lung water measurements during patient management in an intensive care unit. Am Rev Respir Dis 1987; 136:662– 668 5 Bo¨ck J, Lewis FR. Clinical relevance of lung water measurement with the thermal dye dilution technique. J Surg Res 1990; 48:254 –265 6 Mihm FG, Feeley TW, Jamieson SW. Thermal dye double indicator dilution measurement of lung water in man: comparison with gravimetric measurements. Thorax 1987; 42: 72–76 7 Staub N. Clinical use of lung water measurements: report of a workshop. Chest 1986; 90:589 –594 8 Fein A, Grossman RF, Jones JG, et al. Evaluation of transthoracic electrical impedance in the diagnosis of pulmonary edema. Circulation 1979; 60:1156 –1160 9 Spinale FG, Reines HD, Cook MC, et al. Noninvasive estimation of extravascular lung water using bioimpedance. J Surg Res 1989; 47:535–540 10 Korsten HHM, Leusink JA, Spierdijk J, et al. Pulmonary shunting after cardiopulmonary bypass. Eur Heart J 1989; 10(suppl):1–5 11 Jonsson F, Madsen P, Jorgensen LG, et al. Thoracic electrical impedance and fluid balance during aortic surgery. Acta Anaesthesiol Scand 1995; 39:513–517 12 Harris ND, Suggett AJ, Barber DC, et al. Applications of applied potential tomography (APT) in respiratory medicine. Clin Phys Physiol Meas 1987; 8(suppl):155–165 13 Hahn G, Sipinkova I, Baisch F, et al. Changes in thoracic impedance distribution under different ventilatory conditions. Physiol Meas 1995; 16:A161–A173 14 Kunst PW, Vonk Noordegraaf A, Hoekstra O, et al. Ventilation and perfusion imaging by electrical impedance tomography: a comparison with radionuclide scanning. Physiol Meas 1998; 19:481– 490
1702
15 Gattinoni L, Pelosi P, Vitale G, et al. Body position changes redistribute lung computed tomographic density in patients with acute respiratory failure. Anesthesiology 1991; 74:15–23 16 Murray JF, Matthay MA, Luce JM, et al. An expanded definition of the adult respiratory distress syndrome. Am Rev Respir Dis 1988; 138:720 –723 17 Brown BH, Barber DC. Possibilities and problems of realtime imaging of tissue resistivity. Clin Phys Physiol Meas 1988; 9(suppl):121–125 18 Smith RWM, Freeston IL, Brown BH. A real time electrical impedance tomography system for clinical use: design and preliminary results. IEEE Trans Biomed Eng 1995; 40:133– 139 19 Kunst PW, de Vries PMJM, Postmus PE, et al. Evaluation of electrical impedance tomography in the measurement of PEEP-induced changes in lung volume. Chest 1999; 115: 1102–1106 20 Pelosi P, Crotti S, Brazzi L, et al. Computed tomography in adult respiratory distress syndrome: what has it taught us? Eur Respir J 1996; 9:1055–1062 21 McLuckie A. The COLD system of hemodynamic monitoring. Int Care World 1996; 13:24 –28 22 Sivak ED, Wiedemann HP. Clinical measurement of extravascular lung water. Crit Care Clin 1986; 2:511–526 23 Bland JM, Altman DG. Correlation, regression, and repeated data. BMJ 1994; 308:896 24 Gattinoni L, Bombino M, Pelosi P, et al. Lung structure and function in different stages of severe adult respiratory distress syndrome. JAMA 1994; 271:1772–1779 25 Gattinoni L, Pelosi P, Crotti S, et al. Effects of positive end-expiratory pressure on regional distribution of tidal volume and recruitment in adult respiratory distress syndrome. Am J Respir Crit Care Med 1995; 151:1807–1814 26 Bigatello LM, Zapol WM. New approaches to acute lung injury. Br J Anaesth 1996; 77:99 –109 27 Mitchell JP, Schuller D, Calandrino FS, et al. Improved outcome based on fluid management in critically ill patients requiring pulmonary artery catheterization. Am Rev Respir Dis 1992; 145:990 –998 28 Brigham KL, Kariman K, Harris TR, et al. Correlation of oxygenation with vascular permeability-surface area but not with lung water in humans with acute respiratory failure and pulmonary edema. J Clin Invest 1983; 72:339 –349
Clinical Investigations in Critical Care