Automatic Border Detection Identifies Subclinical Anthracycline Cardiotoxicity in Children With Malignancy
Background—Anthracycline drugs for cancer therapy often cause functional myocardial impairment even in relatively low doses. We investigated the left ventricular function in asymptomatic anthracycline-treated children by automatic border detection (ABD) to assess its clinical usefulness for unmasking latent anthracycline-induced myocardial damage.
Methods and Results—Thirty-four children (0.7 to 17.6 years old) during or after anthracycline chemotherapy (26 to 1100 mg/m2) for malignancy (Chemo group) were studied, and 40 children (2.8 to 15.6 years old) without cardiac involvement served as normal control subjects (Control group). All patients underwent complete echocardiographic examination, including M-mode, Doppler, and ABD. Conventional echocardiography disclosed no difference between groups with regard to ejection fraction and the ratio of early to late transmitral flow velocity. In marked contrast, an investigation using ABD revealed that the Chemo group appeared to have some anthracycline-induced myocardial damage. In the apical 4-chamber view, peak filling rate in the Chemo group [2.3±0.4 end-diastolic area (EDA)/s] was significantly lower than that in the Control group (3.1±0.5 EDA/s) (P<0.0001), and time to peak filling rate in the Chemo group (106±31 ms) was clearly prolonged compared with that in the Control group (74±22 ms) (P<0.0001).
Conclusions—Echocardiographic ABD may be a sensitive and useful noninvasive approach for evaluating subclinical anthracycline cardiotoxicity.
Cardiotoxicity induced by anthracycline drugs is very serious, irreversible, and limits the therapeutic potential in many patients undergoing cancer chemotherapy.1 2 Such cardiotoxicity is related to the cumulative dose of these drugs; the risk of congestive heart failure in children has been reported to increase at cumulative doses >550 mg/m2.1 Little has been published on the effects of moderate doses of anthracyclines on myocardial function in clinically asymptomatic children treated for malignant neoplasms.
For early detection of this cardiotoxicity, several noninvasive techniques for evaluation of left ventricular (LV) diastolic function have been devised.3 4 5 6 These techniques do not always improve sensitivity and specificity for detecting this cardiotoxicity.
Automatic border detection (ABD) has been applied clinically in adults for evaluation of volume measurements and LV function.6 Several ABD parameters seem to be highly sensitive for detecting changes in diastolic function.7 8 Thus, we assessed the feasibility of using ABD for detection of subclinical myocardial damage in asymptomatic children treated with moderate doses of anthracyclines compared with conventional echocardiography.
A total of 74 children and infants who had been referred to our hospital for routine echocardiographic evaluation and in whom endocardial visualization of the LV was complete were studied.
Thirty-four patients, 0.7 to 17.6 years old, were studied during or after anthracycline therapy for malignant neoplasms (Chemo group) (Table⇓). None of these patients had received anthracyclines within the previous week and received mediastinal irradiation or had clinically overt signs of heart failure.
The remaining 40 patients, 2.8 to 15.6 years old, were studied as control subjects (Control group); these patients were evaluated in our center because of murmurs or noncardiac chest pain and did not have heart disease according to echocardiographic and physical examinations. There were no significant differences in age or heart rate between the 2 groups, as shown in the Table⇑. In addition, the groups showed no differences in body weight (35±24 versus 27±14 kg) or height (125±24 versus 123±22 cm).
All children were studied with an echocardiographic Hewlett-Packard Sonos 2500 ABD system with a 3.5- or 5.5-MHz phased-array transducer, as described previously.9 After careful adjustment for optimum image definition and identification of the area of interest, automatic tracking of the endocardial border was performed in all subjects in both apical 4-chamber and short-axis views (Figure 1A⇓ and 1C⇓). Calculation of the traced area within the region of interest was performed automatically to provide the AQ waveform shown in Figure 1B⇓ and 1D⇓, assuming the single-plane Simpson’s rule. The peak of this wave shows the end-diastolic area (EDA), and the bottom shows the end-systolic area (ESA). From this wave, fractional area change (FAC) was calculated as (EDA−ESA)/EDA as an index of systolic function. The lower wave was obtained by differentiating the upper wave with respect to time; it shows real-time area change rate. From this wave, peak filling rate (PFR) and time to peak filling rate (TPFR) were obtained. PFR is the maximum area change rate divided by the EDA. TPFR is the time from end-systolic phase to PFR. This system calculates the 5-beat average PFR and TPFR automatically.
Conventional Echocardiographic Methods
For conventional echocardiography, ejection fraction obtained by the Pombo method was calculated as a parameter of LV systolic function. The ratio of early to late peak velocity (E/A) of transmitral inflow recorded from the apical 4-chamber view by pulsed-wave Doppler aimed at the tip of the mitral valve leaflets was measured as a parameter of LV diastolic function.
All studies were interpreted by a single observer who was unaware of clinical details (I.H.). The systolic and diastolic parameters obtained by ABD were compared with those obtained by conventional echocardiography.
All data are expressed as mean±SD. An unpaired Student’st test was performed for comparisons between the Chemo and Control groups. Probability values of <0.05 were considered statistically significant.
Conventional echocardiography demonstrated no difference between the 2 groups in either ejection fraction (0.78±0.07 in the Chemo group versus 0.80±0.07 in the Control group) or E/A ratio (1.7±0.5 versus 1.9±0.5).
In the study with ABD, no significant differences were observed in the FAC between the 2 groups in either the apical 4-chamber (0.34±0.06 versus 0.37±0.06) or short-axis (0.59±0.11 versus 0.62±0.13) view.
In contrast, the PFR of the Chemo group was significantly lower than that of the Control group (2.3±0.4 EDA versus 3.1±0.5 EDA, P<0.0001) in the apical 4-chamber view. In addition, PFR in the apical 4-chamber view was ≥2.5 EDA in the control children but was <2.5 EDA in 22 of the 34 patients in the Chemo group. In the short-axis view, however, no difference was observed in PFR between the 2 groups (Figure 2A⇓). TPFR was significantly prolonged in the Chemo group compared with that in the Control group (106±31 versus 74±22 ms, P<0.0001) in the apical 4-chamber view. This difference was not demonstrable between the 2 groups in the short-axis view (Figure 2B⇓).
This study shows that asymptomatic children treated with moderate doses of anthracyclines for cancer sustain subclinical myocardial damage. We used diastolic ABD parameters rather than systolic parameters in the detection of early anthracycline-induced cardiotoxicity in this study, based on the findings of earlier studies using conventional echocardiography.6 7 Consequently, diastolic PFR and TPFR obtained by ABD successfully unmasked this anthracycline-induced myocardial damage in the early stage, whereas FAC showed systolic function of the LV to be normal in these children despite the use of ABD.
Diastolic parameters were normal in the short-axis view but were decreased in the long-axis view in many children in our study treated with anthracycline. These results suggest that a decrease in LV compliance in the long axis may precede that in the short axis during the course of deterioration of LV diastolic filling, as previously reported in adults with constrictive pericarditis and restrictive cardiomyopathy.10 The short-axis view would be less sensitive to myocardial disease and would show changes only at the short-axis level that has been chosen. We therefore recommend use of LV diastolic parameters in the apical 4-chamber view rather than in the short-axis view for early detection of anthracycline cardiotoxicity in children.
In this study, we used PFR and TPFR as the diastolic parameters, both of which are affected by heart rate and are closely related to body size. To eliminate these influences, we confirmed that there were no significant differences in age, body size, or heart rate between the children treated with anthracyclines and the control subjects. It would take a larger study to establish age- and body surface area–related normal data for children. Because PFR in the apical 4-chamber view was >2.5 EDA in the control children but was <2.5 EDA in many children treated with anthracycline, we consider PFR <2.5 EDA to be abnormal, irrespective of age or body size, and recommend close serial monitoring of cardiac function during and after chemotherapy in those children with PFR <2.5 EDA.
Many asymptomatic children treated with moderate doses of anthracyclines for cancer have subclinical deterioration of diastolic function, especially in the LV long axis. ABD seems to be a useful noninvasive method for early detection of anthracycline-induced myocardial damage and is feasible for small and even debilitated children during and after chemotherapy for cancer. The prognostic importance of subclinical myocardial dysfunction detected by ABD is still uncertain. Intense long-term monitoring of cardiac function in the children in this study is warranted.
We are grateful to Dr David Sahn for critical review of the manuscript, to Dr Hideki Origasa for assistance with the statistical analysis, and to Drs Ken Suzaki, Gyokei Murakami, and Hiromichi Kubota for permission to study patients under their care.
- Received October 22, 1998.
- Revision received February 24, 1999.
- Accepted March 9, 1999.
- Copyright © 1999 by American Heart Association
Sorensen K, Levitt G, Bull C, Chessells J, Sullivan I. Anthracycline dose in childhood acute lymphoblastic leukemia: issues of early survival versus late cardiotoxicity. J Clin Oncol. 1997;15:61–68.
Massing JL, Caillot D, Mouhat T, Schlesser P, Solary E. Early detection of cardiotoxicity of anthracyclines: value of isotopic method. Arch Mal Coeur Vaiss. 1996;89:57–62.
Hashimoto I, Ichida F, Miura M, Okabe T, Shimura S, Uese K, Hamamichi Y, Tsubata S, Miyawaki T, Fukahara K, Murakami A. Evaluation of left ventricular volume using automatic border detection in children: a comparison with conventional off-line echocardiographic quantification. Acta Paediatr Jpn. 1998;40:226–231.