| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Cardiac Imaging |
1 From the Departments of Radiology (A.L., T.H.W., P.R., K.G.A.H., B.H., C.N.H.E.), Internal Medicine I–Cardiology, Angiology, and Pneumology Section (A.C.B.), and Cardiovascular Surgery (S.D., P.M.D.), Charité Medical School, Humboldt University, Berlin, Germany. Received April 4, 2004; revision requested June 17; revision received August 2; accepted September 23. Address correspondence to A.L., Department of Radiology, Massachusetts General Hospital, Harvard Medical School, 55 Fruit St, White Bldg 270, Boston, MA 02114 (e-mail: alexander.lembcke{at}gmx.de).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Institutional review board approval and informed consent were obtained. Volume and flow measurements were performed at electron-beam CT in 219 patients (197 men, 22 women; mean age, 61.5 years ± 10.4 [standard deviation]), of whom 157 had known isolated mitral valve regurgitation. Regurgitation volume was calculated as the difference between left ventricular total and forward stroke volumes. Regurgitation fractions were compared with corresponding echocardiographic grades (grades 0–IV) by using Spearman rank correlation and a weighted 
test. In 22 patients, CT volume and flow measurements were compared with MR results by using intraclass correlation.
RESULTS: Regurgitation fractions at CT correlated well with echocardiographic grading (rank correlation coefficient, rS = 0.82; P < .05). Mean regurgitation fractions for echocardiographic grades 0, I, II, III, and IV were 3.1% ± 6.2, 12.7% ± 9.9, 25.3% ± 12.3, 40.4% ± 11.5, and 55.9% ± 13.7, respectively. The most suitable thresholds for differentiating echocardiographic grades were calculated regurgitation fractions of 6%, 20%, 30%, and 44%; with these thresholds, individual echocardiographic grades were differentiated (grades 0 vs I–IV, 0–I vs II–IV, 0–II vs III–IV, and 0–III vs IV, respectively) with sensitivities of 89%, 87%, 86%, and 93% and specificities of 81%, 87%, 92%, and 91%, respectively. There was perfect agreement in classification of mitral valve insufficiency between electron-beam CT and echocardiography in 134 (61%) patients and a mismatch by one grade in 72 (33%) and by two grades in 13 (6%) ( = 0.84). Intraclass correlation coefficients between CT and MR imaging for total and forward stroke volumes and regurgitation volume and fraction were 0.88, 0.79, 0.93, and 0.89, respectively.
CONCLUSION: Electron-beam CT provides quantitative information on severity of mitral valve regurgitation, but semiquantitative classification of regurgitation showed mismatch between electron-beam CT and Doppler echocardiography by at least one grade in more than one-third of all patients.
© RSNA, 2005
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Nevertheless, only few data are available on the use of electron-beam CT in assessing cardiac valve defect, although the latter represents the second most common indication for cardiac surgery after coronary artery disease. Furthermore, relative insufficiencies of the atrioventricular valves are frequent secondary complications of dilated heart disease and are crucial for the further course of the disease. Thus, the degree of mitral valve regurgitation is an important prognostic factor—mild mitral valve regurgitation does not usually lead to left ventricular remodeling (11,12), whereas left ventricular dysfunction is a serious complication that typically accompanies severe regurgitation (13,14). Moreover, patients with only mild mitral regurgitation have an excellent prognosis (15), while severe regurgitation increases morbidity and mortality (16).
We hypothesized that electron-beam CT can be used to quantify mitral valve incompetence. Thus, the purpose of our study was to prospectively compare regurgitation fractions of the mitral valve calculated with electron-beam CT, by using Doppler echocardiography as the reference standard, and to evaluate the accuracy of electron-beam CT volume and flow measurements compared with results at magnetic resonance (MR) imaging.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Patients who were excluded were those who developed new cardiac events (n = 2), underwent a cardiac intervention (n = 6), had a change in systolic blood pressure of 30 mm Hg or more between the two studies (n = 1) or had concomitant aortic valve incompetence (n = 9), intracardiac shunting (n = 1), or an implanted left ventricular assist device (n = 23). Also excluded were patients who had an inconclusive or questionable echocardiographic diagnosis because of an insufficient sonication window (n = 12), an eccentric regurgitation jet (n = 4), or an implanted metal valve prosthesis (n = 1).
All 219 study patients underwent routine clinical examination, auscultation, and additional transthoracic Doppler echocardiography. Electron-beam CT was performed either before or after Doppler echocardiography, with an interval of 1.9 days ± 2.3 (range, 0–9 days).
The data were acquired prospectively by using standardized imaging protocols.
All examinations (electron-beam CT, echocardiography, and MR imaging) were performed as part of a clinical study on electron-beam CT, which was approved by the institutional review board. All patients gave written informed consent.
Electron-Beam CT
Data acquisition and complete image postprocessing and evaluation (manual tracing and region-of-interest setting) were performed by one of two qualified radiologists (A.L. or T.H.W., both with 4 years of experience in cardiac CT) who knew the patient's underlying disease and major accompanying disorders but was blinded to the results of echocardiographic measurement. All patients were examined with an electron-beam CT scanner (Evolution C-150 XP with software version 12.4; GE Imatron, San Francisco, Calif) at 625 mA and 130 kV. Scanning was performed in the electrocardiographically triggered multisection mode with a section thickness of 8 mm and an acquisition time of 50 msec.
The patients were imaged in the supine position by using a section orientation approximated to the short axis of the heart (table slew to the right by 25° and table tilt to the feet by 17°).
Following acquisition of localizer scans, flow measurement over the ascending aorta was performed after administration of iopromide (Ultravist, iodine content of 370 mg/mL; Schering, Berlin, Germany) at a dose of 0.2 mL per kilogram of body weight. The contrast medium was injected at a rate of 4 mL/sec into a cubital vein, and 30 electrocardiographically triggered scans (two adjacent sections with 15 scans per section) were acquired of the ascending aorta at the level of the pulmonary bifurcation (flow mode). The scanning was triggered as follows: The first scan was acquired simultaneous with the start of contrast medium administration, followed by 14 repetitive scans acquired with an interscan interval of five cardiac cycles. In patients with severe congestive heart failure or tachycardia, the interscan intervals were prolonged on an individual basis. If results of the first flow study were inconclusive, a repeat study was performed after 5 minutes with an increased contrast medium bolus of 0.4 mL per kilogram of body weight and renewed adjustment of the trigger to the patient's presumed individual circulation time if necessary.
For flow measurements, a region of interest with a size between 2 and 4 cm2 was manually placed in the aortic root by the radiologist. The scanner's standard software was used to calculate an attenuation-time curve (Hounsfield units per unit of time) for the ascending aorta from the source data to determine the time of maximum contrast medium concentration (transit time) and cardiac output. The latter was calculated according to the formula of the indicator dilution method as validated in previous studies (17–19): CO = CF · VCM/AUC, where CO is the cardiac output in liters per minute, CF is the calibration factor (5000 HU/370 mg of iodine per milliliter) multiplied by the iodine concentration of the contrast medium, VCM is the volume of the contrast medium in milliliters, and AUC is the area under the attenuation-time curve in Hounsfield units multiplied by seconds. By taking into account the heart rate, the result can then be used to calculate the forward stroke volume per cardiac cycle.
For the subsequent functional study, another 90 mL of contrast medium was injected at a flow rate of 3 mL/sec, and 156 scans were acquired at the time of maximum contrast medium concentration during a single end-expiratory breath hold. These 156 electrocardiographically triggered scans covered the entire left ventricle from apex to base by means of 12 serial acquisitions, each with 13 scans per cardiac cycle (cine mode).
This standard electron-beam CT scanning protocol resulted in a radiation exposure of approximately 7.6 mSv, as reported previously (20).
Cardiac volumes were calculated by using the standard evaluation software implemented on the console of the scanner. This was done by manually tracing the endocardial contours on end-diastolic and end-systolic images (field of view, 18 or 21 mm; matrix, 256 x 256). The software then calculated left ventricular volumes by multiplying the areas by the section thickness and then adding all sections (section summation method). The 4-mm gaps that occur between any two pairs of sections were accounted for by means of interpolation.
The total stroke volume of the left ventricle was the difference between its end-diastolic volume and its end-systolic volume. The regurgitation volume of the mitral valve was calculated as the difference between the total stroke volume of the left ventricle and forward stroke volume measured in the aorta (negative values of the regurgitation volume were defined as zero). The regurgitation fraction (expressed as a percentage) was the proportion of the regurgitation volume relative to the total left ventricular stroke volume.
An example of flow measurement and volume determination is presented in Figure 1.
|
|
In addition, biplane evaluation of the narrowest central flow region of the mitral valve regurgitation jet was performed (absolute width: less than 4.0 mm for grade I; 4.0–5.9 mm for grade II; 6.0–9.0 mm for grade III; and greater than 9.0 mm for grade IV), and, when in doubt, definite classification with the calculation of the effective regurgitant orifice area was performed by using the proximal isovelocity surface area method (effective area size: less than 0.20 cm2 for grade I; 0.20–0.29 cm2 for grade II; 0.30–0.39 cm2 for grade III; and greater than 0.39 cm2 for grade IV).
Intra- and Interobserver Variability of Electron-Beam CT Measurements and Accuracy Compared with MR Imaging Results
The intra- and interobserver variability and the accuracy of volume and flow measurements at electron-beam CT were retrospectively analyzed in all patients who underwent additional MR imaging in the context of their routine clinical work-up. This subset of the entire study population consisted of 22 patients (19 men, three women) ranging in age from 38 to 77 years (mean, 63.3 years ± 9.8). All patients in this subgroup underwent standardized MR imaging with a routine imaging protocol on a 1.5-T imager (Magnetom Vision; Siemens, Erlangen, Germany). Data acquisition and image postprocessing were performed by one of two qualified radiologists (A.L. or T.H.W., both with 2 years of experience in cardiac MR imaging). Left ventricular volume measurements were performed in a sectionwise manner along the doubly angulated short axis from the base to the apex of the heart. Images were acquired during end-expiratory breath holding by using a prospectively triggered two-dimensional fast low-angle shot cine sequence with k-space segmentation and echo sharing (repetition time msec/echo time msec, 80 or 100/4.8; flip angle, 20°; matrix, 126 x 256; field of view, 340–400 mm; section thickness, 10 mm; intersection distance, 0 mm). End-diastolic, end-systolic, and total stroke volumes were calculated with the implemented software (Argus; Siemens), which uses the section summation method. For flow measurement in the ascending aorta, a section was positioned through the cross section exactly at a right angle to the longitudinal axis. Images were acquired during normal breathing by using a two-dimensional fast low-angle shot phase-contrast sequence (24/5.0; flip angle, 30°; matrix, 160 x 256; field of view, 280–360 mm; section thickness, 6 mm; velocity encoding, 250 cm/sec; two signals acquired). The forward stroke volume was measured by using the system's standard software, which allows for generating a flow curve, and by using a region of interest placed over the entire cross section of the ascending aorta.
For calculation of intra- and interobserver variability with regard to electron-beam CT, tracing of endocardial borders of the left ventricle and placing of the regions of interest were performed immediately after the examination by an experienced radiologist (A.L. or T.H.W.). This process was repeated on the same data set after a delay of at least 1 month by the same radiologist and additionally by a second qualified radiologist (C.N.H.E.) with 4 years of experience in cardiac CT. Both were unaware of the results of the initial study.
Descriptive and Statistical Analysis
The relationship between regurgitation fractions determined at electron-beam CT and regurgitation grades established at Doppler echocardiography was presented in a scatterplot and tested by using Spearman rank correlation coefficient. The median values and quartiles of the data are presented in box plots. The Wilcoxon test for unpaired samples was used to test whether the mean values of the regurgitation fractions calculated with electron-beam CT differed from the grades of regurgitation determined with Doppler echocardiography.
To determine how well the five echocardiographic grades of mitral valve incompetence can be distinguished with electron-beam CT, receiver operating characteristic (ROC) curves were generated, the area under the curve for each ROC curve was calculated, and the most suitable threshold for each of the five grades was identified—that is, the value for which the sum of sensitivity and specificity was highest. The positive predictive value (PPV) and negative predictive value (NPV) were also calculated. With these thresholds, the agreement between electron-beam CT and Doppler echocardiography was determined by the calculation of the weighted value.
For the subgroup of patients without valve defects, the agreement between the total stroke volume of the left ventricle and forward stroke volume measured in the aorta was presented in a Bland-Altman scatterplot and tested by using the Student t test for paired samples. The relationship between measurements of total stroke volume and forward stroke volume was assessed by means of linear regression with calculation of the intraclass correlation coefficient, including its 95% confidence interval (CI).
Calculation of the differences and their 95% CIs (ie, the limits of agreement) according to Bland-Altman analysis, the Student t test, and the intraclass correlation coefficient were also used to evaluate the accuracy of volume and flow measurements at electron-beam CT in comparison with the results at MR imaging.
Intra- and interobserver variability were expressed as the absolute difference between two measurements or as the percentage of variability (absolute value of the difference between two measurements divided by the mean of the two measurements) if appropriate.
Significance in all statistical tests was assumed at P < .05, with adjustment of the significance level according to the number of tests performed for those variables that were submitted to multiple testing (Bonferroni correction). Data are expressed as means ± standard deviations.
We used SPSS version 11.0 software (SPSS, Chicago, Ill) for all statistical tests except for the computation of the weighted value, which was calculated by using StatExact (version 5.0; Cytel Software, Cambridge, Mass).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Data from the functional studies performed at electron-beam CT could be evaluated primarily in all 219 patients, whereas flow studies had to be repeated in 46 (21%) patients because the initially acquired data were inadequate to calculate time-attenuation curves owing to inadequate adjustment of the trigger, poor contrast enhancement, motion artifacts, and/or beam hardening.
Opacification of the left ventricle was moderately reduced in some patients because of a markedly prolonged circulation time resulting from severe congestive heart failure. Nevertheless, it was possible in all of these patients to delineate and manually trace the endocardium.
Comparison of Electron-Beam CT and Doppler Echocardiography
Statistical analysis showed a close correlation between the regurgitation fractions determined at electron-beam CT and the regurgitation grades established at Doppler echocardiography (Fig 2a); the Spearman rank correlation coefficient was rs = 0.82 (P < .05).
|
|
The ROC analyses (Fig 3) demonstrated a good discriminatory power for using electron-beam CT to differentiate between the echocardiographic grades of mitral valve insufficiency. The best threshold values were the calculated regurgitation fractions of 6% for grade 0 versus grades I–IV, 20% for grades 0–I versus grades II–IV, 30% for grades 0–II versus grades III–IV, and 44% for grades 0–III versus grade IV. With use of these thresholds, the different echocardiographic grades of mitral valve insufficiency could be differentiated with a sensitivity of 89% (140 of 157), 87% (74 of 85), 86% (42 of 49), and 93% (13 of 14); a specificity of 81% (50 of 62), 87% (116 of 134), 92% (157 of 170), and 91% (186 of 205); a PPV of 92% (140 of 152), 82% (84 of 102), 76% (42 of 55), and 41% (13 of 32); and an NPV of 74% (50 of 67), 91% (116 of 127), 96% (157 of 164), and 99% (186 of 187), respectively.
|
|
|
|
value for agreement between electron-beam CT and Doppler echocardiography in classification of mitral valve regurgitation was = 0.84 (95% CI: 0.79, 0.88).
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Doppler echocardiography is quite accurate and currently regarded as the standard of reference for assessment of mitral valve regurgitation. However, echocardiography also has limitations (24). The most important of these is that its results are reliant on the quality of the patient's sonication window, as well as on the training and experience of the examiner.
Although electron-beam CT has already been used as an alternative method for quantifying mitral valve insufficiency (9,25), results of only two comparative studies with relatively small sample sizes have been presented in the literature to show that electron-beam CT might accurately quantify the mitral valve regurgitation volume and fraction (18,19). In an analogy to MR imaging, and on the basis of the continuity equation, there are two different procedures that might be considered. First, in patients with valvular defects, the regurgitation volume can be calculated from the difference between left ventricular and right ventricular stroke volumes, since these two volumes must be exactly identical in patients with a normal heart (2,25,26). However, this method is potentially associated with an elevated error rate for the following anatomic, physiologic, and technical reasons: more complex geometry of the right ventricle, the plane of the tricuspid is valve more difficult to define than that of the mitral valve, disregard of the bronchial collateral flow, and the difficulty to simultaneously achieve optimal opacification of both ventricles. More important, however, this method is limited by the fact that it can be used only when an isolated valve defect is present. In the majority of patients with dilated or congestive heart disease, the method is invalid because these patients have (relative) insufficiency of both the mitral and tricuspid valve.
For these reasons, determination of the global stroke volume of the left ventricle is combined with measurement of the forward stroke volume in the aorta. This is a method that can be used in nearly all patients except those with simultaneous aortic valve incompetence or intracardiac shunting.
This method relies on volumetry of the left ventricle, which has been validated for electron-beam CT by means of phantom measurements (1) and animal studies (2). Moreover, authors of earlier investigations (5,19) have suggested that there is an excellent correlation between electron-beam CT and MR imaging despite slight differences in absolute measurements. This is an important observation because MR imaging is at present the generally accepted standard of reference for use in measuring left ventricular volumes (27,28).
Regarding flow measurements in the aorta at electron-beam CT, we can likewise rely on the results of earlier studies. Although some authors (29) assume that the indicator dilution method is probably inferior to the section summation method in terms of accuracy, the results reported by other authors who used the indicator dilution method for flow measurements in a phantom (30) and for determination of cardiac stroke volume in animals (17,31) confirm the reliability of this method. Moreover, recently published study results (19) suggest that there is also a good correlation between electron-beam CT flow measurements and MR flow measurements. The latter is regarded as highly accurate and reproducible (32–35) and, thus, is the noninvasive reference standard for determining flow volumes, including cardiac output (36).
Altogether, the results presented here suggest that electron-beam CT is accurate for exclusion of substantial regurgitation defects and suitable for estimating the regurgitation volume and fraction in mitral valve insufficiency, although there remains some subjective diagnostic insecurity involved in this technique because it does not enable direct visualization of the regurgitation jet. The latter represents the major diagnostic limitation of electron-beam CT. However, additional electron-beam CT measurements may be useful in inconclusive cases with disagreement among results from different modalities (echocardiography, left ventriculography, MR imaging) or where these modalities yield inaccurate results—for example, in cases with a limited sonication window, an eccentric regurgitation jet, or a metal valve prosthesis.
Electron-beam CT may be used as an alternative imaging modality to MR imaging for determining the mitral valve regurgitation volume, for measuring the forward stroke volume of the heart, and for quantifying the volumes of both the left and the right ventricle, as well as the myocardial mass, in particular in patients with metal implants (pacemakers or defibrillators) and in patients with severe cardiac failure and higher grade dyspnea who would not tolerate a prolonged MR examination in the flat supine position with repeated breath-hold periods.
The indication for electron-beam CT must be based on strict criteria since it is associated with radiation exposure and intravenous contrast medium administration. However, the quantitative determination of the regurgitation volume does not involve any additional radiation exposure or additional contrast medium administration, since flow measurement is performed during the determination of the transit time, which is routinely performed prior to each functional study. A standard functional study thus yields data sets providing all essential information.
Finally, it must be noted that electron-beam CT is much less widely used than competing modalities such as echocardiography and MR imaging. However, the basic principle of the electron-beam CT protocol presented here seems to be applicable to the emerging multi–detector row CT technology, as well.
Our study was limited by the fact that there is currently no general agreement about the reference standard for diagnosing mitral valve insufficiency. Whereas conventional angiography used to be the generally accepted method of reference in the past, many investigators now prefer Doppler echocardiography. Another drawback was that we followed the general practice and performed only a semiquantitative determination of the severity of mitral valve insufficiency. It is possible, in principle, to quantitatively determine the regurgitation volume and regurgitation fraction. However, quantitative measurement potentially has an even higher level of inaccuracy, as suggested by reports in the literature (37), and we therefore did not consider quantitative measurement a suitable standard of reference.
Finally, it has to be mentioned that comparison between electron-beam CT and MR imaging was performed in a patient subset with a relatively small sample size and with the use of fast low-angle shot sequences. Although these sequences have shown a high level of accuracy and reproducibility for measuring ventricular volumes (28,29), recently developed steady-state free precession sequences have a shorter acquisition time and give a sharper delineation of the endocardial borders, which may result in a higher precision of ventricular volume measurements (38–40).
In summary, electron-beam CT may yield additional data for detecting and quantifying mitral valve insufficiency from a standard cardiac function study. This may be of particular interest in patients with dilated heart disease (especially dilated and ischemic cardiomyopathies, as well as higher grade primary mitral valve insufficiency) who undergo a cardiac function study with electron-beam CT as part of their diagnostic and preoperative work-up or postoperative follow-up. While there is good overall agreement in the assessment of mitral valve incompetence between electron-beam CT and Doppler echocardiography in our study population, the observed discrepancies in individual patients show that all measurement results must always be interpreted by taking into account each patient's individual examination conditions and the potential error rate of the diagnostic modality used.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviations: CI = confidence interval • NPV = negative predictive value • PPV = positive predictive value • ROC = receiver operating characteristic
Authors stated no financial relationship to disclose.
Author contributions: Guarantors of integrity of entire study, A.L., B.H., C.N.H.E.; study concepts and design, A.L., B.H., C.N.H.E.; literature research, A.L., A.C.B.; clinical studies, A.L., A.C.B., S.D., P.M.D., T.H.W., C.N.H.E.; data acquisition, A.L., A.C.B., S.D., P.M.D., T.H.W., C.N.H.E.; data analysis/interpretation, all authors; statistical analysis, A.L.; manuscript preparation and editing, A.L.; manuscript definition of intellectual content, revision/review, and final version approval, all authors
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  P. M. Dohmen, A. Lembcke, S. Holinski, D. Kivelitz, J. P. Braun, A. Pruss, and W. Konertz Mid-Term Clinical Results Using a Tissue-Engineered Pulmonary Valve to Reconstruct the Right Ventricular Outflow Tract During the Ross Procedure Ann. Thorac. Surg., September 1, 2007; 84(3): 729 - 736. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
) (outliers are defined as values more than 1.5 box lengths away from the end of the box), and extremums (
) (extremums are defined as values more than three box lengths away from the end of the box). Horizontal lines in the box plots delineate thresholds of regurgitation fraction that best separate echocardiographic grades. There is a close relationship between the CT regurgitation fractions and Doppler echocardiographic regurgitation grades, with Spearman rank correlation coefficient of rs = 0.82 (P < .05).
