Recognition of Acute Cardiac Allograft Rejection From Serial Integrated Backscatter Analyses in Human Orthotopic Heart Transplant Recipients
Comparison With Conventional Echocardiography
Background Previous studies showed that moderate and severe acute cardiac rejection (AR) but not mild AR is associated with significant myocardial acoustic changes. This study examines whether serial measurements of end-diastolic two-dimensional integrated backscatter (2D-IB) enhance the diagnostic potential of ultrasonic tissue analysis in AR.
Methods and Results Serial endomyocardial biopsies, conventional echocardiograms, and parasternal long-axis radiofrequency signals for determination of posterior wall and septal 2D-IB were performed in 52 transplant patients. Histology showed no AR in 155 biopsy samples, AR grade 1A in 25, AR grade 1B/2 in 27, and AR grade 3A/3B in 13. Whereas no significant 2D-IB changes occurred between AR-free studies and during AR grade 1A, posterior wall and septal 2D-IB increased during AR grade 1B/2 from −47.80±4.36 to −42.97±5.11 dB and from −36.72±7.45 to −32.52±7.98 dB (P<.001 and P<.05, respectively) and during AR grade 3A/3B from −47.96±4.74 to −38.25±5.32 dB and from −37.92±5.87 to −31.01±4.62 dB (P<.001 and P<.01, respectively). Changes in posterior wall and septal 2D-IB were greater during AR grade 3A/3B than during AR grade 1B/2 (P<.01 and P<.05). Increases of 1.5 dB in posterior wall or septal 2D-IB indicated AR grades ≥1B with sensitivities of 88% and 83% and specificities of 89% and 85%; posterior wall and septal 2D-IB increases of 5.5 and 3.8 dB identified AR grades ≥3A with sensitivities of 92% and 79% and specificities of 90% and 84%. Although a weak inverse correlation between posterior wall and septal 2D-IB changes and posterior wall and septal thickening (r=.41 and r=.39, both P<.001) and fractional diameter shortening (r=.35, P<.001) was found, significant 2D-IB increases also occurred in some rejecting patients with unaltered contraction.
Conclusions Increases in end-diastolic posterior wall and septal 2D-IB in serial studies permit reliable identification not only of moderate and severe AR but also of mild AR. Because 2D-IB increases significantly more in AR with myocyte damage than without such damage, an estimate of AR severity appears feasible. Significant myocardial acoustic changes during AR may occur independently of changes in contractile performance.
Histologically, acute cardiac allograft rejection is characterized by myocardial edema, cellular infiltrates, and, in more severe cases, myocyte damage and interstitial hemorrhage.1 The main criterion for the usefulness of any diagnostic tool for rejection surveillance is its ability to rapidly and reliably identify rejection-induced myocardial changes. Although endomyocardial biopsy has some inherent limitations2 3 and bears a small risk for the patient,4 5 6 histology has remained the acknowledged “gold standard” for rejection diagnosis because no other method has been shown to fulfill this requirement consistently.
Echocardiographic tissue analysis aims to identify myocardial pathology from changes of the acoustic properties of the heart muscle.7 Data obtained from human orthotopic heart transplant recipients indicate that moderate acute rejection significantly alters the magnitude of the systolic-diastolic variation of myocardial IB.8 Several animal experiments involving different heterotopic transplant models used video densitometry to show increases in end-diastolic myocardial echo amplitude during acute rejection.9 10 11 However, only in a study from our laboratory involving daily ultrasonic and histological allograft assessment were the animals maintained on standard immunosuppression similar to that used in humans11 ; the study demonstrated that under these conditions, moderate and severe rejection altered end-diastolic myocardial echo amplitude significantly, whereas milder rejection grades did not induce consistent acoustic changes. Thus, lack of sensitivity limits the diagnostic applicability of video densitometry in humans, in whom detection of mild diffuse rejection is also desirable in view of a potential risk of subsequent progression in rejection severity.12 13 14
The present prospective clinical study was devised to determine whether serial measurements of end-diastolic 2D-IB from unprocessed myocardial RF signals would enhance the potential of echocardiographic tissue analysis as a diagnostic tool for detection of acute cardiac rejection. The main objectives of the study were (1) to examine in what way end-diastolic myocardial 2D-IB is affected by acute cardiac rejection and how these changes compare with simultaneously obtained conventional echocardiographic parameters and (2) to determine the sensitivity and specificity of the end-diastolic myocardial 2D-IB changes for identifying acute cardiac allograft rejection and, if possible, estimating its severity. To achieve these goals, intraindividual reproducibility of end-diastolic 2D-IB in serial rejection-free examinations and interindividual comparability of the measurements were also investigated.
Seventy-eight consecutive adult orthotopic heart transplant recipients undergoing routine postoperative echocardiographic monitoring were considered for the present study. All patients had to have normal left ventricular dimensions, wall thickness, and systolic function; posttransplant arterial hypertension had to be well controlled by antihypertensive therapy with ACE inhibitors, diuretics, or calcium antagonists throughout the observation period. Acute cardiac rejection had to be excluded by endomyocardial biopsy at the time of the first echocardiographic examination. Patients were either enrolled before their first rejection episode or after complete histological resolution of previous rejection. Twelve of the total group of consecutive patients fulfilling the inclusion criteria were excluded for inadequate image quality and 14 were excluded because only one biopsy-correlated echocardiographic examination was performed during the observation period. Thus, the total study population consisted of 52 patients. There were 13 females and 39 males with a mean age of 41±13 years. Patients entered the study 8 to 2180 days (median, 66 days) after orthotopic transplantation and were followed up for an observation period of 37 to 461 days (median, 172 days).
Standard Immunosuppression and Rejection Therapy
All patients received standard immunosuppression including prophylactic antithymocyte globulin/antilympohcyte globulin for the initial 2 postoperative weeks, cyclosporin A (daily dosage adjusted on the basis of plasma trough levels), prednisone (5 to 7.5 mg/d from the second postoperative month, higher doses in the first postoperative month), and azathioprine (1 to 2 mg/d). Episodes of diffuse mild rejection (within the first 6 months of heart transplantation) or moderate and severe rejection (irrespective of the time elapsed since heart transplantation) were treated with parenteral high-dose methylprednisolone (500 to 1000 mg/d for 3 consecutive days). Persistent acute rejection despite repeat parenteral high-dose steroids in one case was treated with oral methotrexate (7.5 mg every other day for 4 weeks).
Monitoring for acute allograft rejection was based on the histological results of serial right ventricular endomyocardial biopsies performed according to the protocol for routine rejection surveillance established at our institution: at weekly intervals during the 1st month (starting on the 6th or 7th postoperative day), at 2- to 3-week intervals between the 2nd and 4th month, at 6-week intervals until the 7th month, and at 12-week intervals until the 24th month. Thereafter, biopsies were performed twice a year. Additional biopsies for control of rejection therapy were performed as required. Each time, a minimum of four tissue samples was obtained. The biopsy samples were graded according to the modified Billingham criteria15 : grade 0, no rejection; grade 1A, focal (perivascular or interstitial) infiltrate without myocyte damage (regardless of the time elapsed since transplantation, grade 1A rejection is neither treated nor controlled by biopsy); grade 1B, diffuse but sparse infiltrate without myocyte damage (grade 1B corresponds to mild diffuse rejection); grade 2, one focus only with aggressive infiltrate and/or myocyte damage; grade 3A, multifocal aggressive infiltrates and/or myocyte damage (grade 3A corresponds to moderate rejection); and grade 3B, diffuse inflammatory process with myocyte necrosis (diffuse myocyte necrosis always indicates severe rejection). Biopsy samples were graded by one experienced pathologist (Dr Weiss).
M-mode and two-dimensional echocardiograms were performed with the use of a Toshiba SSH 160 electronic sector scanner (Toshiba Medical Systems Europe) by two of the investigators (Drs Stempfle and Kru¨ger) within 24 hours of the endomyocardial biopsy. In all studies, the same 2.5-MHz transducer was used. Examinations were performed with the patient in a left lateral decubitus position and were recorded on videotape. The transducer was placed at the left parasternal border and adjusted to obtain an optimal image of the long-axis view of the left ventricle. Transducer position and gain settings were documented for each individual during the first examination and reproduced for all subsequent studies. Maximal transmission power was always used. Before each follow-up examination, each patient's first recording was reviewed to ensure precise reproduction of the imaging plane. M-mode echocardiograms of the left ventricle and an ECG were recorded at a paper speed of 50 mm/s.
Left ventricular end-diastolic and end-systolic cavity dimensions and septal and posterior wall thicknesses were measured from the M-mode recordings as recommended by the American Society of Echocardiography.16 Fractional diameter shortening and wall thickening were calculated in a standard manner. Data from three cardiac cycles were averaged. All measurements were performed without knowledge of the corresponding biopsy histology.
Acquisition and Analysis of RF Signals
In addition to conventional echocardiography, unprocessed RF signals were obtained regularly. A block diagram of the data acquisition and analysis system is shown in Fig 1⇓. The system was equipped with a prototype interface for signal-level adjustment and synchronization, allowing for real-time digitization of data in other than video formats.17 18 Maximal amplitude resolution was 12 bits, and maximal sampling frequency was 30 MHz. The raw ultrasound signal was obtained from the Toshiba SSH 160 electronic sector scanner before logarithmic amplification, demodulation, and scan conversion; the ultrasound signal was unaffected by time-gain compensation and scanner-integrated postprocessing. The dynamic range of the data was ≈100 dB.
In the present study, an amplitude resolution of 8 bits and a sampling frequency of 20 MHz were chosen. A digitally controlled amplifier (256 steps, maximal amplification of 40 dB) was used for linear amplification of the RF data; this device was adjusted to facilitate optimal use of the input dynamic range of the analog-to-digital converter (48 dB) for quantitative assessment of the myocardial backscatter from the proximal segments of the ventricular septum and posterior wall. With respiration suspended, RF data corresponding to entire frames of standardized left parasternal long-axis views were digitized on-line into random access memory. ECG triggering was used, and only end-diastolic RF data were captured.
RF images as obtained in this study are rectangular, because they are not scan converted (Fig 2⇓, top). Visual assessment is usually difficult. Thus, the analysis program implemented in the image processing system displayed simultaneously both the original RF data and the scan-converted image (Fig 2⇓, bottom). With use of a digitizer, ROIs are marked in the scan-converted image, in which the endocardial borders are more clearly delineated. The system then automatically selects the corresponding areas of the RF image for quantitative analysis.
ROIs that were 1500 to 2400 pixels in size were defined within the myocardium of the septum and posterior wall, with endocardial and epicardial borders carefully excluded. The sizes of the regions were maximized individually to include the full thickness of the ventricular wall. In all patients, printouts showing raw as well as scan-converted images and demonstrating clearly the size and position of the regions were archived. During all subsequent studies, these printouts were used to reproduce the regions precisely for each individual subject.
Time-averaged 2D-IB as a measure of the total energy of the returning ultrasound signal was calculated in these regions by use of the equation2D-IB|<|=|>||<|\int_|<|A|>||>|S(a) dawhere S(a) represents the current backscatter signal, A represents an area within the myocardial echo image with dimensions x and h, and da represents the infinitesimal area element. 2D-IB was expressed in decibels, by use of the equation 2D-IB(dB)=−20∗log (REF/2D-IB). REF represents an internal reference, which corresponds to the IB value of a perfect reflector. Again, values from three cardiac cycles were averaged; analysis was performed without knowledge of the biopsy result.
Because, as previously noted, ultrasonic transmission power has a tendency to decrease with increasing transducer age, acoustic power was monitored at regular time intervals during the study by use of a tissue equivalent phantom with known transmission and attenuation characteristics (RMI model 821; Gammex-RMI GmbH). Because no significant changes were observed throughout the observation period, this ensured near-constant transmission power of the transducer and therefore comparability of the consecutive 2D-IB measurements in individual subjects.
Data are expressed as mean±SD. For each individual, changes occurring during histologically proven acute rejection episodes were compared with the last preceding rejection-free measurement (No AR). Data were grouped according to rejection severity; consecutive rejection-free measurements were also compared (No AR versus No AR, No AR versus acute rejection grade 1A, No AR versus acute rejection grade 1B/2, and No AR versus acute rejection grade 3A/3B). The two-tailed Student's t test was used to compare the paired data sets; ANOVA was used to test differences between changes observed in a subgroup of patients suffering from repeat rejection episodes with different rejection severity. Relations between two variables were studied by use of simple linear regression analysis. Intraindividual variability of consecutive rejection-free studies was determined as mean difference and mean percent difference of the measurements. Sensitivity (true-positives divided by true-positives plus false-negatives) and specificity (true-negatives divided by true-negatives plus false-positives) of rejection-induced changes in end-diastolic 2D-IB as an index of acute cardiac rejection were determined by constructing ROC curves, for which various cutoff points may be selected from continuous scales of values to adjust the sensitivity and specificity of the test to conform to clinical needs.19
A total of 246 routine endomyocardial biopsies were performed in the study patients within 24 hours of echocardiography. The biopsy material was considered inadequate in 26 instances; thus, 220 biopsies were screened for the presence of acute allograft rejection. Histology showed rejection grade 0 in 155 instances, grade 1A in 25, grade 1B or 2 in 27, and grade 3A or 3B in 13. In 45 patients, two or more consecutive grade 0 biopsy samples were obtained.
Conventional Echocardiographic Parameter and End-Diastolic 2D-IB Measurements Versus Histopathology
Values (mean±SD) of all investigated parameters are listed in Table 1⇓. In addition to the mean values of consecutive rejection-free studies, measurements during histologically proven acute rejection were compared with values obtained from the same patients during the last preceding rejection-free study. Data are grouped according to rejection severity. In accordance with the inclusion criteria, left ventricular dimensions, wall thickness, and function were normal in all patients during rejection-free studies. Mean values for left ventricular end-diastolic dimension and wall thickness remained unchanged, whereas fractional diameter shortening tended to decrease with increasing rejection severity. During rejection episodes grade 1B/2 and grade 3A/3B, posterior wall and septal thickening were significantly decreased compared with each patient's preceding rejection-free study (P<.001 and P<.05 for posterior wall thickening and P<.005 and P<.05 for septal thickening during rejection grades 1B/2 and 3A/3B, respectively). In a subgroup of eight patients who suffered from mild diffuse (grade 1B) as well as moderate or severe rejection episodes (grade 3A or 3B) during the observation period, posterior wall and septal thickening decreased significantly from 77.0±22.0% and 46.8±11.8% to 60.7±12.2% and 33.6±8.7%, respectively, during rejection grade 1B (both P<.01). During rejection grade 3A/3B, an additional small decrease occurred to 58.9±11.8% and 33.4±8.7%, respectively. However, values obtained during mild versus moderate or severe rejection were not statistically different.
Fig 3⇓ shows individual posterior wall (top) and septal end-diastolic 2D-IB values (bottom). Fig 3⇓ depicts consecutive rejection-free measurements and compares measurements obtained during acute rejection of increasing severity with the preceding rejection-free values. Corresponding mean values are listed in Table 1⇑. Whereas end-diastolic 2D-IB did not change significantly between rejection-free measurements and during rejection grade 1A, it increased during most grade 1B or 2 rejection episodes and in all grade 3A or 3B rejection episodes. This resulted in significant increases in the respective mean values (P<.001 for increases in posterior wall end-diastolic 2D-IB during grades 1B/2 and 3A/3B, respectively, and P<.05 and P<.005 for the corresponding septal values); as illustrated in Fig 4⇓, posterior wall and septal 2D-IB increases were significantly more pronounced with moderate or severe rejection than with mild rejection in the eight patients who experienced both mild and more severe rejection. In this subgroup, posterior wall 2D-IB increased from −48.0±4.5 to −44.5±6.0 dB during rejection grade 1B (P<.01) and to −39.0±5.3 dB during grades 3A/3B (P<.01 for both grade 0 versus grades 3A/3B and grade 1B versus grades 3A/3B). Fig 5⇓ illustrates rejection-induced myocardial acoustic changes in two original RF images obtained in a standardized fashion from the same patient. A visible change in myocardial backscatter intensity occurred during rejection grade 3A compared with the rejection-free image. Quantification revealed a rejection-induced 10.7-dB increase in posterior wall 2D-IB.
Reproducibility of rejection-free measurements was assessed from the 45 pairs of consecutive studies in which acute cardiac rejection was histologically excluded. Mean absolute and percent differences of all echocardiographic measurements are given in Table 2⇓. In patients with three or more consecutive rejection-free studies (n=23), mean absolute and percent differences for posterior wall and septal end-diastolic 2D-IB were in the same range (between 1.20±1.80 and 1.47±1.28 dB [2.78±4.62% and 3.26±2.89%] for posterior wall end-diastolic 2D-IB and 0.47±0.28 and 1.25±1.01 dB [1.24±0.84% and 4.62±4.13%] for septal end-diastolic 2D-IB) between the second and third, third and fourth, and fourth and fifth examinations.
Fig 6⇓ depicts ROC curves describing the sensitivity and specificity of posterior wall (top) and septal end-diastolic 2D-IB changes (bottom) as indexes for acute cardiac rejection. A histological rejection grade of either ≥1B or ≥3A was chosen as the criterion for constructing the ROC curves. In both the posterior wall and the septal graphs, the points of interception of the sensitivity and specificity curves for rejection grades ≥1B were found at 1.5 dB; a +1.5-dB change in posterior wall and septal end-diastolic 2D-IB identified acute cardiac rejection ≥1B with sensitivities of 88% and 83% and specificities of 89% and 85%, respectively. The corresponding numbers of true-positive and false-positive and true-negative and false-negative values are given in Table 3⇓. The points of interception of the posterior wall and septal sensitivity and specificity curves for rejection grades ≥3A were found at 5.5 and 3.8 dB, respectively. Using these thresholds, end-diastolic 2D-IB measurements identified acute cardiac rejection ≥3A with sensitivities of 92% and 79% and specificities of 90% and 84%, respectively.
Sensitivity and specificity of conventional echocardiographic parameters were also calculated; each time, 95% confidence limits corresponding to two SDs of the mean absolute difference of consecutive rejection-free measurements were used as a threshold. Changes exceeding the threshold were considered suggestive of rejection. All tested parameters had a satisfactory specificity (range, 84% to 87%) but a low sensitivity (posterior wall thickening, 43%; septal thickening, 40%; fractional shortening, 55%). If a significant change in any of these parameters was used as a rejection criterion, sensitivity increased to 80% but specificity decreased to 68%.
Relation Between Changes in End-Diastolic 2D-IB Measurements and Changes in Conventional Echocardiographic Parameters
Fig 7⇓ compares changes in posterior wall and septal end-diastolic 2D-IB between consecutive studies with corresponding changes in posterior wall (top), septal thickening (middle), and left ventricular fractional diameter shortening (bottom). There was a rather weak but significant inverse correlation between changes in posterior wall and septal end-diastolic 2D-IB and systolic wall thickening (n=110, r=.41, P<.001 for the posterior wall and r=.39, P<.001 for the ventricular septum). End-diastolic myocardial 2D-IB was also inversely correlated with left ventricular fractional diameter shortening, but this relation was even weaker (n=110, r=.35, P<.001). Fig 7⇓ illustrates that in many patients, rejection grades 1B/2 or 3A/3B were not only associated with an increase in end-diastolic 2D-IB but also with a decrease in regional and global left ventricular function. However, in some cases, no change or very small changes in left ventricular wall thickening and fractional diameter shortening occurred despite marked increases in myocardial end-diastolic 2D-IB. In addition, Fig 7⇓ illustrates the significant variability in left ventricular function parameters in consecutive rejection-free studies and shows that no directional changes occurred during rejection grade 1A.
In this study, we performed echocardiographic tissue analysis in cardiac allografts with and without biopsy-proven acute rejection using end-diastolic posterior wall and septal 2D-IB for serial assessment of myocardial acoustic properties. Experimental and clinical studies indicate that backscatter is altered significantly in various myocardial pathologies, including acute ischemia, infarction and reperfusion,20 21 22 23 scar tissue and fibrosis,24 25 26 and abnormalities in tissue water content and blood flow.27 In an animal study from our laboratory,11 daily quantitative analyses of end-diastolic myocardial echo amplitudes and transmural biopsies were performed in 12 dogs with a cervical heterotopic transplant treated with standard immunosuppression. With increasing rejection severity, a concomitant progressive increase in myocardial video intensity was observed, which suggested that rejection-induced myocardial edema, cellular infiltrates, and, in more advanced stages, myocyte damage and hemorrhage also lead to alterations of the myocardial acoustic properties. As in previous experimental studies,9 10 however, video densitometry revealed significant increases in myocardial echo amplitude only with moderate or severe rejection, whereas mild rejection episodes were not reliably identified. As a consequence, the present investigation in humans was based on end-diastolic myocardial 2D-IB measurements, which are derived from the unprocessed RF signal. Whereas data compression is performed by logarithmic amplification in video images, unprocessed RF signals undergo linear amplification only; therefore, it was hypothesized that subtle acoustic changes, such as may be associated with early rejection stages, might be more readily detectable from these data. In addition, the RF signals obtained in the present study were independent of time-gain compensation, postprocessing, and signal interpolation during scan conversion.17 Thus, it was also hypothesized that serial examinations of standardized echocardiographic cross sections in individual patients would produce strictly comparable results, whereas interindividual comparisons would remain problematic because of variable signal attenuation in different subjects. Because no published data on intraindividual reproducibility or interindividual variability of end-diastolic myocardial 2D-IB measurements are available, however, these hypotheses were also tested in the present clinical study investigating the diagnostic potential of end-diastolic 2D-IB changes in acute cardiac allograft rejection.
Intraindividual Reproducibility and Interindividual Comparability of Myocardial End-Diastolic 2D-IB Measurements and Conventional Echocardiographic Parameters
As illustrated in Fig 3⇑, intraindividual reproducibility of the posterior wall as well as septal 2D-IB measurements proved very satisfactory in rejection-free examinations. Thus, if intercostal space, ultrasonic imaging plane, and size and position of the ROIs are kept constant, serial end-diastolic 2D-IB measurements as performed in the present study represent a robust and stable diagnostic tool. If one assumes no alteration in the state of the tissue in consecutive rejection-free studies, it is likely that smaller changes in myocardial 2D-IB, which were observed in some patients, were due to technical factors such as suboptimal reproduction of the imaging plane or the position of the ROI. In view of the angle dependency of ultrasound28 and the potential regional heterogeneities of the myocardial structure, reproducibility would very likely decrease if differing imaging planes and ROIs were used. Strictly standardized acquisition and quantification of the RF data are thus mandatory if the end-diastolic 2D-IB measurements are to be used for diagnostic purposes.
Fig 3⇑ also demonstrates a broad interindividual variability of the measurements that exceeds the maximal intraindividual 2D-IB changes observed during acute rejection. This proves that in contrast to the cyclic IB variation, which is widely used for comparisons among different subjects,7 such end-diastolic 2D-IB measurements can only be used for serial intraindividual monitoring. Methods to correct for losses in signal intensity caused by attenuation within tissue layers and blood interposed between the transducer and a specific ROI have been developed previously in vitro29 and have been applied in vivo in various myocardial pathologies.26 30 31 Alternatively, the individual signal intensity of the pericardium24 30 31 or the blood adjacent to the investigated tissue area32 has been used to “normalize” myocardial backscatter. However, each of these approaches aiming at meaningful standardization of the 2D-IB measurements seems to have some inherent uncertainty, because signal attenuation in blood and soft tissue may differ in vitro and in vivo and is associated with several technical sources of error29 ; on the other hand, saturation or very low intensity of the reference measurements may preclude true calibration of the myocardial 2D-IB. Thus, intraindividual monitoring of end-diastolic 2D-IB measurements appears at present to be the most appropriate application for this tissue parameter.
Changes of Myocardial End-Diastolic 2D-IB and of Conventional Echocardiographic Parameters During Acute Cardiac Allograft Rejection
The present study confirms once more that acute cardiac allograft rejection leads to an alteration of the myocardial acoustic properties.8 9 10 11 In accordance with our previous experimental results11 in dogs receiving maintenance immunosuppression similar to that used in the present study, a significant increase in myocardial 2D-IB was demonstrated during moderate or severe rejection. In addition, however, significant changes in posterior wall and septal end-diastolic 2D-IB were observed during rejection grades 1B or 2, facilitating even the noninvasive identification of rejection, without myocyte damage in most instances. As a previous comparative study from our laboratory showed,18 this seems to indicate a superior ability of end-diastolic 2D-IB measurements compared with video densitometry to recognize more subtle changes in the myocardial acoustic properties. On the other hand, the lack of changes in end-diastolic 2D-IB during rejection grade 1A appears to justify the widespread clinical interpretation of this histological phenomenon as a nonsignificant and sometimes nonspecific finding, which necessitates neither intensified immunosuppression nor a control biopsy.
Numerous previous investigations11 33 34 35 36 37 38 demonstrated that acute cardiac rejection may be associated with impairment of regional and global myocardial contractile function and with an increase in left ventricular mass, although in cyclosporine-treated patients, measurements usually remain within the normal range.39 Most studies agree, however, that single conventional echocardiographic parameters cannot identify rejection with satisfactory sensitivity and specificity11 33 34 35 because of the rather heterogenous morphological and functional consequences of rejection episodes, even of those of the same histological grade. A possible explanation of this phenomenon might be differing degrees of concomitant vascular rejection, which may go undetected if immunofluorescent staining of the histological specimens is not routinely performed in addition to hematoxylin-eosin staining; recent evidence suggests that the response of left ventricular mass and function to vascular compared with histologically evident cellular rejection may be different.38
Fig 7⇑ demonstrates that in the present study, regional and global left ventricular function decreased in many rejecting heart transplant recipients while end-diastolic myocardial 2D-IB increased. However, in a number of patients who also showed significant 2D-IB increases, contractile performance remained unchanged despite histologically proven moderate or severe rejection. This shows that rejection-induced acoustic changes can occur independently of changes in myocardial function. In a smaller population of heart transplant recipients, Masuyama et al8 observed significant decreases in posterior wall and septal cycle-dependent variation in IB during moderate acute allograft rejection, while regional function remained unaltered statistically, although it tended to decrease. The present study supports the hypothesis advanced by Masuyama et al that although myocardial contractile performance is one of the contributors to alterations in IB variation during acute cardiac rejection, it may not be the sole determinant. It seems plausible that rejection-induced changes in structure, elasticity, or other properties of the heart muscle leading to an increase in myocardial end-diastolic 2D-IB would also influence cyclic IB variation. It would have been interesting to measure both parameters simultaneously to determine their relationship, but this was not feasible for technical reasons. Thus, it remains uncertain to what extent the diagnostic ability of cyclic IB changes in acute cardiac allograft rejection depends on concomitant changes in myocardial contractile performance.
Sensitivity and Specificity of End-Diastolic Myocardial 2D-IB Changes as Indexes of Acute Cardiac Allograft Rejection; Estimation of Rejection Severity
The rejection-induced changes in posterior wall as well as septal backscatter intensity were significantly more pronounced in rejection episodes with myocyte damage than in episodes without myocyte damage, thus facilitating for the first time a noninvasive estimate of rejection severity. In contrast, changes in conventional echocardiographic parameters neither permitted discrimination between rejection grades 1B/2 and 3A/3B nor facilitated identification of the presence of rejection with comparable sensitivity and specificity. Previous studies33 34 35 suggest that investigation of a greater number of different variables, including presence or absence of pericardial effusion, visual assessment of myocardial echo brightness, and Doppler indexes of diastolic function, might have improved the diagnostic yield. However, it was not the purpose of the present study to evaluate more than the standard echocardiographic measurements in comparison with the quantitative myocardial 2D-IB analysis.
Sensitivity and specificity of end-diastolic 2D-IB measurements as indexes of acute rejection were better in the posterior wall than in the septum. This observation corresponds to the results of Masuyama et al,8 who found the diagnostic value of the cyclic variation of posterior wall IB superior to that of the septum. A possible explanation for this finding might be the better reproducibility of the posterior wall 2D-IB measurements, as shown in Table 2⇑. Although end-diastolic posterior wall 2D-IB measurements seem to indicate rejection more accurately than septal measurements, it seems appropriate for clinical purposes to obtain both values, because concordant changes will further strengthen diagnostic reliability.
As indicated in Table 3⇑, a discrepancy between histological rejection grade and end-diastolic posterior wall and septal 2D-IB values was found in several instances. For rejection grade 1B/2 (threshold 1.5 dB), one possible explanation for false-positive or false-negative results is the variability of 2D-IB measurements even in consecutive rejection-free studies, which, as shown in Table 2⇑, is in the range of 1 to 2 dB. Furthermore, histological and echocardiographic evaluation of the myocardium are not performed in the same regions of the heart; biopsy specimens are routinely sampled from the right ventricular septum or apex, whereas 2D-IB measurements are obtained from the left ventricular walls. Our own data in animals11 and experimental results by others40 suggest that the rejection process may not always affect the right and left ventricles to the same extent and that rejection may vary in severity throughout the left ventricle. This may explain why several patients showed pronounced increases in posterior wall or septal 2D-IB, although right ventricular histology demonstrated a rejection grade of only 1B or 2. Such measurements were classified as false-positive for rejection grade 3A/3B, but as illustrated by Fig 7⇑, they were correct with regard to the diagnosis of rejection; no measurement with biopsy-excluded rejection or acute rejection grade 1A fell into this false-positive group. On the other hand, some patients with histological rejection grade 3A/3B showed only moderate increases in posterior wall or septal 2D-IB. Fig 7⇑ illustrates, however, that in all patients with rejection grades 3A or 3B, an increase in posterior wall and septal end-diastolic 2D-IB by >1.5 dB suggested the presence of rejection. With the 2D-IB values classified as false-negative for rejection grades 3A/3B, the diagnosis of rejection, therefore, was not missed, although rejection severity was not correctly recognized. Thus, according to the results of the present study, increases in end-diastolic myocardial 2D-IB values obtained under strictly standardized conditions represent a reliable index for moderate and severe but also for mild rejection; rejection severity as estimated from the 2D-IB measurements, however, does not always correspond precisely to histology. Personal experience and results by others35 suggest that in clinical practice, the decision about intensified immunosuppressive therapy may be aided by additional conventional echocardiographic measurements, because spontaneous resolution of rejection seems less likely with significant increases in left ventricular mass or functional impairment.
Several limitations need to be mentioned. First, the image acquisition and processing system used in this study is not commercially available. In addition, the memory-intensive storage and time-consuming analysis of the RF data would at present preclude widespread use of end-diastolic myocardial 2D-IB measurements as a clinical tool for rejection surveillance in heart transplant recipients. However, this study has proven once more the considerable diagnostic potential of this parameter; with the growing interest of ultrasound manufacturers in RF imaging, it is hoped that future commercial equipment will facilitate more cost-effective and user-friendly determination of end-diastolic 2D-IB from unprocessed RF data. Furthermore, adequate ultrasonic images could not be obtained in 12 (15.4%) of the initial 78 patients; Masuyama et al8 reported a similar percentage of nonechogenic heart transplant recipients. Thus, it seems predictable that 10% to 20% of the cardiac transplant population will not profit from this noninvasive diagnostic modality for technical reasons. Finally, as in most transplant centers, immunohistochemistry was not routinely performed in the present study. Precise information on concomitant vascular rejection might have enhanced understanding of the sometimes discordant changes in ventricular myocardial backscatter and function.
This study confirms that acute cardiac allograft rejection is associated with significant alterations of the myocardial acoustic properties. Increases in end-diastolic myocardial 2D-IB in serial studies, in which each patient is used as his or her own control, permit reliable identification not only of moderate and severe but also of mild rejection. Because the 2D-IB increase was significantly more pronounced in rejection grades with myocyte damage than in those without myocyte damage in patients with repeat rejection episodes, the data suggest that a rough noninvasive estimate of rejection severity might be feasible. Although the intraindividual reproducibility of rejection-free myocardial 2D-IB is good, if measurements are strictly standardized, the broad variability of the rejection-free values between different subjects precludes separation of patients with and without rejection by use of interindividual 2D-IB comparisons. Although group data showed a significant decrease in regional pump function during rejection, myocardial contractile performance remained unchanged in several patients with moderate or more severe rejection grades. Comparison of conventional echocardiographic parameters and myocardial 2D-IB measurements showed that significant rejection-induced acoustic changes may occur independently of changes in myocardial function. With development of appropriate commercial equipment, end-diastolic myocardial 2D-IB measurements may become a clinically useful diagnostic tool for noninvasive serial rejection surveillance in the future.
Selected Abbreviations and Acronyms
|2D-IB||=||two-dimensional integrated backscatter|
|ROC||=||receiver operating characteristic|
|ROI||=||region of interest|
This study was supported by the Wilhelm Sander-Foundation (grant No. 86.015.3).
Reprint requests to Prof Christiane E. Angermann, MD, Medizinische Klinik, Kardiologische Abteilung, Klinikum Innenstadt, Universita¨t Mu¨nchen, Ziemssenstraße 1, D-80336 Mu¨nchen, Germany.
- Received February 7, 1996.
- Revision received August 8, 1996.
- Accepted August 19, 1996.
- Copyright © 1997 by American Heart Association
Billingham ME. Diagnosis of cardiac rejection by endomyocardial biopsy. J Heart Transplant. 1982;1:25-30.
Shanes JG, Ghali J, Billingham ME, Ferrans VJ, Fenoglio JJ, Edwards WD, Tsai CC, Saffitz JE, Isner J, Furtner S, Subramanian R. Interobserver variability in the pathologic interpretation of endomyocardial biopsy results. Circulation. 1987;75:401-405.
Ladowski JS, Griffith BP, Hardesty RL, Armitage JM, Bernstein RL, Bahnson HT. Endomyocardial biopsy in heart transplantation: false positives caused by previous biopsy. Transplant Proc. 1988;10:760-761.
Skorton DJ, Miller JG, Wickline S, Barzilai B, Collins SM, Perez JE. Ultrasonic characterization of cardiovascular tissue. In: Marcus ML, Skorton DJ, Schelbert HR, Wolf GL, eds. Cardiac Imaging. Philadelphia, Pa: WB Saunders Co; 1991:538-556.
Masuyama T, Valantine HA, Gibbons R, Schnittger I, Popp RL. Serial measurement of integrated ultrasonic backscatter in human cardiac allografts for the recognition of acute rejection. Circulation. 1990;81:829-839.
Stempfle HU, Angermann CE, Kraml P, Schu¨tz A, Kemkes BM, Theisen K. Serial changes during acute cardiac allograft rejection: quantitative ultrasound tissue analysis versus myocardial histologic findings. J Am Coll Cardiol. 1993;1:310-317.
Yoseum SA, Tazelaar HD. Histologic evaluation of endomyocardial and lung biopsies in heart and heart-lung transplant recipients. In: Kapoor AS, Laks H, Schroeder JS, Yacoub MH, eds. Cardiomyopathies and Heart-Lung Transplantation. New York, NY: McGraw-Hill Publishing Co; 1991:335-358.
Sahn DJ, DeMaria A, Kisslo J, Weyman A, for the Committee on M-mode Standardization of the American Society of Echocardiography. Recommendations regarding quantitation of M-mode echocardiography: results of a survey of echocardiographic measurements. Circulation. 1978;58:1072-1083.
Drewello R, Stempfle HU, Kru¨ger TM, Nassau K, Angermann CE. High resolution online digitization of unprocessed radiofrequency (RF) data for echocardiographic tissue characterization. In: IEEE Proceedings; Second European Conference on Engineering and Medicine; 1993:611-614.
Angermann CE, Kru¨ger TM, Junge R, Drewello R, Stempfle HU, Spes CH, Theisen K. Intravenous albunex during transesophageal echocardiography: quantitative assessment by videodensitometry and integrated backscatter analysis from unprocessed radiofrequency signals. J Am Soc Echocardiogr. 1995;8:839-853.
Haendchen RV, Ong K, Fishbein MC, Zwehl W, Meerbaum S, Corday E. Early differentiation of infarcted and noninfarcted reperfused myocardium in dogs by quantitative analysis of regional myocardial echo amplitude. Circ Res. 1985;57:718-728.
Angermann CE, Brandl CB, Stempfle HU, Drewello R, Kru¨ger TM, Theisen K. Stress induced transient myocardial ischemia is associated with an increase in myocardial echointensity, which precedes regional myocardial contraction abnormalities. Circulation. 1993;88(suppl I):I-607. Abstract.
Hoyt RH, Collins SM, Skorton DJ, Ericksen EE, Conyers D. Assessment of fibrosis in infarcted human hearts by analysis of ultrasonic backscatter. Circulation. 1985;71:740-744.
Picano E, Pelosi G, Marzilli M, Lattanzi F, Benassi A, Landini L, L'Abbate A. In vivo quantitative ultrasonic evaluation of myocardial fibrosis in humans. Circulation. 1990;81:58-64.
Mimbs JW, Bauwens D, Cohen RD, O'Donnell M, Miller JG, Sobel BE. Effects of myocardial ischemia on quantitative ultrasonic backscatter and identification of responsible determinants. Circ Res. 1981;49:89-96.
Skorton DJ, Collins SM, Woskoff SD, Bean JA, Melton HE. Range and azimuth-dependent variability of image texture in two-dimensional echocardiograms. Circulation. 1983;68:834-840.
Lattanzi F, Di Bello V, Picano E, Caputo MT, Talarico L, Di Muro C, Landini L, Santoro G, Giusti C, Distante A. Normal ultrasonic myocardial reflectivity in athletes with increased left ventricular mass: a tissue characterization study. Circulation. 1992;85:1828-1834.
Naito J, Masuyama T, Tanouchi J, Mano T, Kondo H, Yamamoto K, Nagano R, Hori M, Inoue M, Kamada T. Analysis of transmural trend of myocardial integrated ultrasound backscatter for differentiation of hypertrophic cardiomyopathy and ventricular hypertrophy due to hypertension. J Am Coll Cardiol. 1994;24:17-24.
Ciliberto GR, Cataldo G, Cipriani M, Mascarello M, Faletra F, Gronda E, De Maria R, Mauri L, Pezzano A. Echographic assessment of cardiac allograft rejection. Eur Heart J. 1989;10:400-408.
Hansen DE, Daughters GT II, Alderman EL, Stinson EB, Baldwin JC, Miller DG. Effect of acute human cardiac allograft rejection on left ventricular systolic torsion and diastolic recoil measured by intramyocardial markers. Circulation. 1987;76:998-1008.
Sagar KB, Hastillo A, Wolfgang TC, Lower RR, Hess ML. Left ventricular mass by M-mode echocardiography in cardiac transplant patients with acute rejection. Circulation. 1981;64:216-220.
Angermann CE, Spes CH. Echocardiography in heart transplantation. In: Roelandt JRTC, Sutherland GR, Iliceto S, Linker DT, eds. European Textbook of Cardiac Ultrasound. Edinburgh, Scotland: Churchill Livingstone; 1993:565-578.
Haverich A, Scott WC, Dawkins KD, Billingham ME, Jamieson SW. Asymmetric pattern of rejection following orthotopic cardiac transplantation in primates. J Heart Transplant. 1984;4:280-285.