Noninvasive Assessment of Right Ventricular Diastolic Function by Electrical Impedance Tomography

Noninvasive Assessment of Right Ventricular Diastolic Function by Electrical Impedance Tomography* Anton Vonk Noordegraaf, MD; Theo]. C. Faes, PhD; Andre Janse, MSc; Johan T. Marcus, PhD; Jean G. F. Bronzwaer, MD; Pieter E. Postmus, MD, FCCP; and Peter M .]. M. de Vries, MD

Study objectives: Electrical impedance tomography (EIT) offers the possibility to study blood volume changes within the right atrium during the cardiac cycle. The aim of this study was to determine the applicability of EIT in the assessment of right ventricular diastolic function in COPD. Design: By means of region of interest analysis, impedance changes within the right atrium during the cardiac cycle were plotted as a function of time. As a diastolic index of the right ventricle, the right atrium emptying volume (RAEV), defined as the ratio between the volume change during the rapid filling phase relative to the total ventricular filling volume, was calculated. In a first study, the validity of the EIT method was assessed by comparison of the RAEV measured by EIT and MRI in a group of eight patients with severe COPD and seven control subjects. A second study was undertaken to assess the relation between RAEV and pulmonary artery pressure in a group of 27 patients measured by right-sided heart catheterization. Results: The correlation coefficient between RAEV measured with MRI and EIT was 0. 78. The difference between RAEV measured by MRI and EITwas 8.3±15.7% (mean±SD) for the control subjects and 3.5± 10.9% for the COPD patients. RAEV values measured by EIT and MRI were larger in the control group (47.1±7.6%) compared with the patient group (38.1±10.4%). There was a clear nonlinear relationship between RAEV and the pulmonary artery pressure (y=315 x- 0 ·6 \ r=0.83, p<0.001). Conclusion: Our results indicate that RAEV measured by EIT is a useful noninvasive and inexpensive method for assessing right ventricular diastolic function in COPD patients. (CHEST 1997; 111:1222-28) Key words: catheterization; diastole; electrical impedance; magnetic resonance imaging; pulmonary hypertension; tomography Abbreviations: EIT=electrical impedance tomography; mPAP = mean pulmonary artery pressure; PAP = pulmonary artery pressure; PH=pulmonary hypertension; RAEV=right atrium emptying volume; ROC=receiver operating characteristic; ROI=region of interest; RV = right ventricle

T he development of a pulmonary hypertension

(PH), leading to a clinical cor pulmonale, is a frequently observed complication in patients with COPD . The primary mechanisms for these complications are vasoconstriction and changes in structure

*From the Departments of Pulmonary Medicine (Drs.Vonk Noordegraaf, Janse, Postmus, and de Vries), Medical Physics and Informatics (Drs. Faes and Marcus), and Cardiology (Dr. Bronzwaer), Institute for Cardiovascular Research (ICAR-VU), Academic Hospital Vrije Universiteit, Amsterdam, the Netherlands. Supported by Glaxo Wellcome pic, the Netherlands. Manuscript received August 9, 1996; revision accepted December 30. Reprint requests: Peter M.]. M. de Vries, MD, PhD, Academic Hospital Vrije Unit;ersiteit, Department of Pulmonary Medicine, PO Box 7057, 1007MB Amsterdam, the Netherlands 1222

of the microarterial bed causing a reduction of the pulmonary vascular bed.1 Patients with cor pulmonale have low right ventricular (RV) distensibility, which is an important factor determining the activity and prognosis of the patient. 2 Accurate monitoring of reductions in RV distensibility will help in providing a proper management of the cardiovascular system throughout the clinical course of COPD . Among the methods presently available, echoDoppler has gained the widest acceptance in the study of RV diastolic dysfunction. 3 The principle of this technique is based on the characteristic biphasic appearance of trans-tricuspid flow velocities. The ratio between the early diastolic, or "passive" flow, and late diastolic, or "active" flow, is used as a measure of ventricular distensibility. However, it is Clinical Investigations

difficult to obtain satisfactory tricuspid flow patterns in COPD patients by means of echo-Doppler. Other t echniques that have the possibility to assess the diastolic function of the RV are MRI 4 and multigated equilibrium radionuclide ventriculography with Krypton-8lm (8 1mKr). 5 In contrast with echo-Doppler, these techniques make it possible to calculate the volume patterns over the tricuspid valves. Takao et al5 showed that the right atrium emptying volume (RAEV), defined as the ratio between the volume change during the rapid filling phase relative to the total ventricular filling volume, not only provides information about the RV diastolic function , but is also a useful test to predict pulmonary artery pressure (PAP).5 However, MRI and multigated equilibrium radionuclide ventriculography with 8 1 mKr are expensive, making them less suitable for routine monitoring in clinical practice. ECG-gated electrical impedance tomography (EIT) is a n ew impedance imaging technique that produces cross-sectional images of the e lecbical impedance distribution within the plane spanned by the e lectrodes. Applied at the thorax, the EIT method offers the possibility to localize visually ventricular and right atrium volume changes during the cardiac cycle.6-9 The method makes it possible to detect the three clinical phases of the diastole: (1)the rapid filling phase; (2) the slow filling phase; and (3) the atrial contraction.9 From these data, it is possible to calculate RAEV by means of EIT (RAEVEIT). In comparison with other noninvasive techniques used for the estimation of RV diastolic function , EIT could have several advantages. First, the method is inexpensive. Second, the method is easy to transport and applicable in a way similar to ECG. The prese nt study was unde rtaken to determine the applicability of EIT in the assessment of RV diastolic function in COPD . In a first study, the validity of the EIT method to measure diastolic function in COPD patients , was assessed by comparing EIT (RAEVEIT ) with MRI (RAEVMRI ) results in a group of patients with seve re COPD and control subjects . A second study was unde rtaken to assess the predictive value of RAEVEIT in the diagnosis of PH .

MATERIALS AND METHODS Patients Seven healthy male subjects ( mean age. 63.8±7.4 years ) without any s ign of cardiac or respiratory diseases and eight COPD patients (seven m en na d one woman; mean age, 63.4±9.6 years) were included for th e MRI protocol. Functional cri teri a for COPD were as follows: increased total lung capacity (> 110% of predicted ); increased residual volume with a esidual r volume/

total lung capacity ratio of > 0.40, associated with an obstructive spirographic pattern measured b y the ratio of FEV1 to vital capacity of -s0.40. Twenty-seven patients scheduled for a diagnostic right-sided heart catheterization were included in the second part of the study. They w ere in clinically s table condition and showed sinus rhythm. Reasons for right-sided h eart catheterization were development of dyspnea after previous infarction (18 patients ), evaluation of severe valvular disease (seven patients), one patient with prima1y PH, and one patient \vith chronic pulmonary e mbolism. The research protocol was approved by the institutional human ethics committee. All patients gave their informed consent. Study Design

In the first p art of the study, we tested the valiclity of the EIT method to measure RV diastolic function in COPD patients. Therefore, we compared the volume, of the rapid filling phase expressed as a percentage of stroke volume, both calculated by MRI (RAEVMRI) with RAEVEIT. As EIT and MRI measurements could not be performed simultaneously, EIT measurerried out immediately after the MRI session with ments were ca the patient in the same position . In the se cond part of the study, the relationship between RAEVEIT and PAP was studied. PAP was measured in a group of cardiologic patients scheduled for right-sided heart catheterization, as this procedure is not perform ed routinely in COPD patients in our institute. EIT measure ments were carried out 1 h after right-sided heart catheterization under the same baseline conditions as during the catheterization; the patient remained in the supine position. Heart rate measured during catheterization did not vary more than five beats compared with EIT measurements. To determine the interobserver v ariation, EIT data were processed by two independent observers. Both obse1vers were blinded for the PAP data measured during right-sided heart catheterization.

EIT EIT measure me nts were performed with the same equipment (Sheffie ld Applied Pote ntial Tomograph ; DAS-01P Portable D ata Acq uisition Syste m; Mark I; IBEES; Sheffi eld, England). •o.u Measure ments were made using an array of 16 equidistantly spaced ele c trodes around the thorax. In this study, the e lectrodes we re attached in an oblique circular plane defin ed at the le vel of th e ictus cordis anteriorly and 10 em highe r poste riorly. 9 A current source, which gene rates a harmonic current (50 kH z, 5 mA peak-peak), was used to measure impedance. D ata collection was synchronized \vith the Rwave of the ECG. Two hundred cardiac cycles were a veraged to obtain one complete data set . One data set contained 30 images spaced 40 ms in time. Each image visualizes the impedance distribution \vithin the thorax. The total EIT protocol lasted 15 min. The subjects were breathing normally during the measurements. To gain visual and quantitative information of the volume changes in the atria, the follo\vin g steps were necessary. First , the EIT required the definition of a r eference data set. All images were reconstructed r elative to this data set. In our study, we defined the refe rence data set asthe average value of the first five frames (end-diastole). In the produced images, the color blue indicated an increase and the color r ed a de crease of impedance relative to end-diastole . Since impedance and blood volume a re inversely proportional, a blood volume decrease in the ventricles during systole causing a n impedance increase was visualized as a blue color. At the same time, a blood volume increase in the atria CHEST I 111 I 5 I MAY, 1997

1223

EIT

MRI

time (s) ,.-.. 0.000 ~

$

d)

bJ)

0.16

0.32

0.48

0.64

0.80

-0.004

~

..d

u -0.008

d)

J d)

c.c

.§

-0.012 -0.016

systole

III diastole

FIGURE 1. Top right: an MRI image, for orientation made in the same oblique plane as th e EIT image. Th e letters RA, RV, LV, and LA are used to indicate th e right atrium and ventricle and the left at rium and ventricle, respectively. Top left: EIT image made at e nd-systole represe nted as the difference to an end-diastolic refe rence . During systole, blood volum e decreases in th e ventricles, appearing as an increase in impedance (blue), and blood volum e increases in th e atria, appearing as a decrease in impedance ( red ). The right atrial region in th e image is demarcated. Bottom: impedance changes during th e cardiac cycle within th e right atrial region as a function of tim e. The systolic part and th e three phases of diastole are indicated in th e curve . RAEVEIT was calculated accord ing to th e fo llowing equ ation: RAEVEIT= {(A-B )/A}X lOO. AU=arbitrary un it.

1224

Clinical Investigations

causing an impedance decrease was visualized as a red color (Fig 1, left upper image). The MRI image (Fig 1, right upper image) made in the same oblique plane as the EIT image showed the anatomic position of different cardiac compartments. From this MRI image, it is clear that the bilobular blue spot represented the ventricles and the encircled red spot represented the right atrium. 9 To quantify the impedance change within the right atrial region, region of interest (ROI) analysis was performed. An ROI was defined around the right atrial region, including all pixels with impedance values of -0.005 or less to the region by drawing a contour line automatically (white demarcation in EIT image). The computer calculated the impedance changes within the ROI over the whole sequence of images. The average pixel value of this ROI was plotted as a function of time, to show the impedance change in the right atrial region during the cardiac cycle (Fig 1, bottom). The average pixel value had no unit because it is dimensionless as a consequence of the reconstruction algorithm based on normalized differences. Therefore, the change of the average pixel value in the sequence during the cardiac cycle relative to end-diastole was expressed as an arbitrary unit. We regarded the lowest impedance (A) as the end-systolic impedance. B indicates the impedance at the middle of diastole (the diastolic phase between the rapid filling phase and the atrial contraction). We defined the RAEV (RAEVErT) as the percentage of right atrium outflow in the first half of diastole, relative to the total filling volume ((A-B)/A)XlOO%. Right-Sided Heart Catheterization Right-sided heart catheterization was performed "vith the patient in the supine position with the use a of 7F single-lumen balloon-tipped catheter (Arrow International; Reading, Pa). The mean PAP (mPAP) was measured with the patient normally breathing. Pulmonary hypertension was defined as mPAP >20 mmHg.

Statistical Analysis Values were expressed as the mean±SD. Correlation analysis was used to analyze the relation between RAEVEIT and RAEVMRI for the healthy subjects and COPD patients. The difference and the SD between both methods were calculated and tested (Mann-Whitney) for the patient and control group separately. Furthermore, the patient data were tested against the control subjects by means of the Mann-\"'hitney test for comparing data from two independent groups. Regression analysis was used to analyze the relation between RAEVEIT (calculated as the mean of the hvo repeated EIT measurements) and mPAP. In addition, receiver operating characteristic (ROC) analysis was performed to determine the optimal cutoff level for RAEVEIT to diagnose pulmonary hypertension (mPAP >20 mm Hg) with EIT. To study the interobserver variation, we used the Pearson's correlation coefficientY A p value <0.05 was considered significant.

RESULTS

Figure 2 shows a scatter plot of RAEVEIT and RAEVMRI results measured in healthy subjects and COPD patients as well as the line of unity. The correlation coefficient between both methods was 0. 78. The difference between both methods was 8.3±15.7% for the healthy subjects and 3.5±10.9% for the COPD patients; both values did not differ significantly from zero (p>0.8). In the control group, RAEV was 47.1±7.6% for EIT and 55.4±14.9% for MRI (p>0.3), and in the COPD group, 38.1±10.4% and 41.6± 16.6% (p>0.3), respectively. The RAEV

MRI MRI measurements were performed on a 1-T whole-body system (Impact Expert; Siemens; Erlangen, Germany), using a phased-array body coil. The flow into and out of the RV was evaluated by phase-contrast velocity quantification 4 ·12 The RV diastolic filling was evaluated by quantifying the flow into the RV, across the tricuspid valves. The planning of the image plane for the tricuspid flow was similar to the planning for the mitral flow as applied by Fujita et al. 13 MRI phase-contrast velocity quantification (temporal resolution=28 ms; field of view=(300 mm)2; matrix, 230X256) was then applied for the tricuspid flow, v.>ith through-plane velocity encoding of 75 cm/s. At least 35 phases in the cardiac cycle were imaged. The tricuspid flow curve was evaluated as follows: phase unwrapping was first performed if necessary. Then, in each separate time phase of the velocity images, the cross-sectional area of the valvular annulus was delineated by hand, in order to account for translations with respect to the image plane. The spatial mean velocity measured in this area was plotted against time. The volume flow curve was obtained by multiplying the spatial mean velocity times the cross-sectional area. Integrating the volume flow curve over the time interval of rapid filling yielded rapid filling volume. The RV stroke volume was obtained by a similar flow quantification applied for the main pulmonary artery, with the image plane adjusted orthogonally to this artery just before the pulmonary bifurcation. Velocity encoding was 150 cm/s and the volume flow curve was integrated over systole, thus yielding stroke volume. Finally, the rapid filling volume was expressed as percentage of the RV stroke volume (RAEVMRI).

80

••

70

,

60 -

40

•

~ ~

•

•

30 20

• •

•

10 -

0

••

•

50

~ > w

• •

L_~L_~L_~L_~L_~--~L_~L_--

0

10

20

30

40

50

60

70

80

RAEVMRI FIGURE 2. Scatterplot between RAEVEIT and RAEVMRI results and the line of unity. The correlation coefficient between both methods was 0.78. The difference between both methods was 8.3± 15.7% for the healthy subjects and 3.5± 10.9% for the COPD patients; both differences did not differ significantly from zero (p>0.8). CHEST I 111 I 5 I MAY, 1997

1225

values, measured by MRI and EIT, were not significantly larger in the control group compared with the patient group (p=0.2). RAEVEIT could not be determined in two cardiologic patients as it was impossible to separate the right atrium region from the surrounding lung region in the EIT image. Both patients were excluded from further analysis. The correlation coefficient between RAEVEIT results obtained from two observers was 0.91. The relationship between RAEVEIT and the PAP of the 23 cardiologic patients is shown in Fig 3, top. The relation is clearly nonlinear; a multiplicative model y=315 x- o. 64 fitted the data properly with

normally distributed residuals (Fig 3, bottom). The correlation coefficient between RAEVEIT and mPAP was 0.83 (p20 mm Hg) was present in 10 patients (42% of the cases). The ROC analysis on decreased RAEVEIT in PH is shown in Figure 4. In case the EIT is used as a screening test for the diagnosis of PH, a 100% sensitivity will be preferable. At a cutoff level of RAEVEIT of 45%, a 100% sensitivity and a 79% specificity were reached in the diagnosis of PH by means of EIT.

DISCUSSION

Diastolic abnormalities are one of the earliest manifestations of cardiac dysfunction.15 Earlier studies have already shown that the study of the diastolic filling of the RV is clinically useful in evaluating the diastolic function of the heart in COPD. 3 •5 •16 Figure l shows that it is possible to study the volume changes in the right atrium by means of ROI analysis. As blood volume changes in the atria and the lungs are in the same direction, impedance changes related to the pulmonary perfusion must be separated to distinguish blood volume changes in the atria. The separation between both regions was based on the difference in magnitude of the impedance changes between both organs, as impedance changes in the right atrium due to the cardiac contraction are larger than impedance changes due to pulmonary perfusion. The pixel value of -0.005 was chosen as, to our experience, a good separation between blood volume changes in the right atrium and in the lung was obtained in this way. Only in two cardiologic patients was it impossible to separate the

80

• •

70

y = 315 x -O 64, r =- 0.83 p < 0.001

•

•• • • • • •• •• •• ••

60

50

20

•

• • •

10

mPAP (mmHg) 40

30

•

20

• • •• •

10

100

• •

90

•

80

- ------- , \ \ \

'-

70

• •••• •• •• •

-10

60 50 40

-20

30 20

-30

-40

10 t__~_L_~__L~_L__~J_~_L~--'-'---'

0

10

20

30

40

50

60

70

rnPAP (mrnHg) FIGURE 3. Top: scatter diagram showing the relation between RAEVEIT and rnPAP (y=315 x- 064 , r=0.83, p
o~--~--~~--~~--~~--~-L----~

20

30

40

50

60

70

80

RAEVEIT

FIGURE 4. ROC curve showing the calculated sensitivity and specificity in the diagnosis of PH for different RAEVEIT values. At a cutoff level of 45% RAEVEIT, a 100% sensitivity and a 79% specificity were reached. Clinical Investigations

right atrial region from the surrounding lung region and left atrium in the EIT image by means of automatic contour drawing. According to radionuclide angiography and echoDoppler cardiography studies, we used RAEV as an index of RV diastolic function.3 · 16 To exclude subjective interpreting factors , we defined the RAEV (RAEVEIT) as the volume change during the first half of diastole (which resembles the rapid filling volume) relative to the total ventricular filling volume (Fig 1). This might explain the high interobserver agreement as indicated by the correlation coefficient of 0.91 between the results of both observers. Figure 2 shows a good agreement between RAEVEIT and RAEVMRI measurements in the control and COPD patients. The mean value for the difference between both methods did not differ significantly and was not larger for COPD patients than for control subjects. This indicates that changes in thoracic cavity and respiratory movements occurring in COPD patients did not limit the usefulness of EIT. Furthermore, both the RAEVEIT and RAEVMRI were reduced in the COPD group in comparison with the control group, although not significantly, suggesting an altered diastolic RV function in COPD. A reason for the nonsignificant difference between control subjects and patients might be the small number of subjects included in the study. Probably, RV wall hypertrophy and pressure overload play an important factor in the altered diastolic filling of the RV. Therefore, the relation between RAEVEIT and mP AP was assessed. The presence of a clear relationship between RAEVEIT and mPAP, measured directly by rightsided heart catheterization, suggests that this technique is useful in estimating PAP noninvasively. The correlation coefficient between RAEVEIT and mPAP is in the same order in comparison with other noninvasive techniques as Doppler echocardiographf·17 and 81 mKr-equilibrium radionuclide ventriculography. Since this study was performed to assess the applicability of EIT in patients with COPD, we used the definition of PH defined by Oswald-Mammoser et al.l 8 They defined PH as mPAP >20 mm Hg based on their findings that only 20% of the emphysematous patients show a resting mPAP >20 mm Hg. The results of the ROC analysis suggest that RAEVEIT can be a sensitive and specific predictor in the diagnosis of PH . Whereas EIT is much less expensive than other a~proaches to assess RV diastolic function (MRI and 8 mKr-equilibrium radionuclide ventriculography), this method makes it possible to assess RV diastolic function in large groups of COPD patients. This not

only offers the possibility to monitor therapeutic effects on RV diastolic function in COPD patients but also to diagnose cardiac dysfunction in an early stage of the disease.

CONCLUSION In conclusion, our results indicate that EIT might be a useful noninvasive and inexpensive method for assessing RV diastolic function in COPD patients. Furthermore, the relationship between RAEVEIT and PAP indicates that the level of RAEVEIT allows estimation of PAP and might be a useful tool in the diagnosis of PH.

REFERENCES 2 3 4 5

6 7

8

9

10 11 12

13

14 15

Reid LM. Structure and function in pulmonary hypertension: new perceptions. Chest 1986; 89:279-88 Fishman AP. Chronic cor pulmonale. Am Rev Respir Dis 1976; 60:91-100 Marangoni S, Scalvini S, Schena M, et al. Right ventticular diastolic function in chronic obstructive lung disease. Eur Respir J 1992; 5:438-43 Mostbeck GH, Hartiala JJ, Foster E, et al. Right ventricular diastolic filling: evaluation with velocity-encoded cine MRI. J Comput Assist Tomogr 1993; 17:245-52 Takao M, Miyahara Y, Hinboku H, et al. Noninvasive assessment of right heart function by 81 MKr equilibrium radionuclide ventriculography in chronic pulmonary disease. Chest 1996; 109:67-72 Eyiiboglu BM, Brown BH , Barber DC. In vivo imaging of cardiac related impedance changes. IEEE Eng Med Bioi 1989; 15:39-45 McArdle FJ. Investigation of imaged conductivity changes due to volume changes of the heart. In: McArdle FJ, ed. Investigation on carcliosynchronous images of the heart and head using applied potential tomography (thesis ). Sheffield: Univ of Sheffield, 1992; 70-82 Eyiiboglu BM , Brown BH, Barber DC, e t al. Localization of cardiac related impedance changes in the thorax. Clin Phys Physiol Meas 1987; 8A:167-73 Vonk Noordegraaf A, Faes TJC, Janse A, et al. Improvement of cardiac imaging in electtical impedance tomography by means of a new electrode configuration. Physiol Meas 1996; 17:179-88 Brown BH, Seagar AD. The Sheffield data collection system. Clin Phys Physiol Meas 1987; 8A:91-7 Smith RWM, Freeston IL, Brown BH. Areal time electrical impedance tomography system for clinical use- design and preliminary results. IEEE Biomed 1995; BME-31:133-40 Van Rossum AC, Sprenger M, Visser FC, et al. An in dvo validation of quantitative blood flow imaging in arteries and veins using magnetic resonance phase-shift techniques . Eur Heart J 1991; 12:117-26 Fujita N, Chazouilleres AF, Hartiala JJ, e t al. Quantification of mitral regurgitation by velocity-encoded cine nuclear magnetic resonance imaging. J Am Coli Cardiol 1994; 23: 951-58 Snedecor KS, Gibson DC. Statistical methods. Ames, Iowa: Iowa State Univ Press, 1989; 183-86 King KS, Gibson DC. Impairment of diastolic function by CHEST/111/5/MAY, 1997

1227

shortened filling period in severe left ventricular disease. Br Heatt J 1989; 62:246-52 16 Berger HJ, Matthay RA, Loke J, et a!. Assessment of cardiac performance \vith quantitative radionuclide angiography: right ventricular ejection fraction with reference to findings in chronic obstructive pulmonary disease. Am J Cardiol 1978; 41:897-905

1228

17 Eysmann SB, Palevsky HI, Reichek N, eta!. Echo/Doppler and hemodynamic correlates of vasodilator responsiveness in primary pulmonary hypertension. Chest 1991; 99: 1066-71 18 Oswald-Mammoser M, Apprill M, Bachez P, et al. Pulmonaty hemody11amics in chronic obstructive pulmonary disease of the emphysematous type. Respiration 1991; 58:304-10

Clinical Investigations