Background
The long-term sequelae of Kawasaki disease (KD) are based on the coronary complications. Because KD causes generalized vasculitis, with documented aneurysms in the femoral, iliac, renal, axillary, and brachial arteries, the aim of this study was to assess the biophysical properties of the aorta (BPA) after KD. The BPA are biometric measurements representing vascular structural and dynamic changes in response to cardiac work.
Methods
Anthropometric and echocardiographic measurements of the aorta in a series of patients with KD were compared with those of healthy subjects. The BPA were calculated noninvasively by extrapolating previously validated equations that were conceived for invasive measurements. Because BPA vary with body habitus, control subjects were used to normalize BPA parameters for height to compute BPA Z -score equations.
Results
Between June 2007 and February 2010, BPA were recorded in 57 patients with KD >1 year after the onset of the disease, 45 without and 12 with coronary artery sequelae. The mean intervals between the acute onset of KD and enrollment were 10.0 ± 5.0 and 5.8 ± 4.5 years for patients with and without coronary artery sequelae, respectively ( P = .008). Patients with KD with coronary artery sequelae had significantly altered Z scores of aortic diameter modulation, Peterson’s elastic modulus, and β stiffness index ( P = .001–.016). Patients with KD without coronary artery sequelae also exhibited altered elasticity, stiffness, and pulse-wave velocity ( P = .001−.026).
Conclusions
Altered BPA after KD are detectible despite apparent resolution of acute vasculitis. Future directions toward determining multilevel and multilayer vascular impact, including vascular autonomous homeostasis, require thorough investigation.
Kawasaki disease (KD) is a systemic vasculitis of unknown etiology in infants and young children that targets coronary arteries (CAs) as well as small or intermediate arteries (arterioles), such as the iliac arteries and the extravisceral portion of the renal, mesenteric, hepatic, or pancreatic arteries. During the acute stage, the widespread vascular endothelial damage begins with degeneration in the endothelium and the adventitia, and inflammatory cells then infiltrate centrally to the media. In later stages, the inflammatory reaction regresses, leaving perivascular fibrosis or scar tissue. The prognosis of KD depends largely on the extent of CA complications, possibly leading to myocardial infarction and death during the convalescent stage in patients with coronary aneurysmal lesions or to progressive coronary stenosis later on. Although coronary function after KD may be altered even in patients without coronary aneurysms, the long-term effects on peripheral endothelial dysfunction remain controversial. On the other hand, the assessment of the biophysical properties of the aorta (BPA), a set of biometric measurements representative of vascular structural and dynamic changes in response to cardiac work, is a useful method to evaluate vascular function in adults. The BPA are also believed to be powerful predictors of cardiovascular disease. However, very few studies have explored the potential of assessment of the BPA in children. In this work, we sought to evaluate the BPA in a series of children after KD on the basis of the fact that KD causes pan vasculitis, including the large arteries. We hypothesized that KD affects BPA in patients with severe CA vasculitis (such as aneurysms) and, to a lesser extent, in those without severe CA vasculitis. Finally, to correct for variation of BPA with growth, we established normative values in healthy children because it is suggested that such parameters vary with somatic growth and age.
Methods
Groups and Subjects
This work was undertaken after ethical approval was obtained. Patients were children with histories of KD followed at our institutions who were randomly selected from the pediatric cardiology follow-up clinic. Clinical history of KD was captured from medical records. Exclusion criteria of subjects with KD were age <1 year, <1 year of follow-up, and use of cardiovascular medications (with the exception of aminosalicylic acid and anticoagulants). Healthy controls were children evaluated in the pediatric cardiology clinic for heart murmurs who had normal results on clinical evaluation, electrocardiography, and echocardiography. Clinical history and anthropometric and echocardiographic data were recorded. Patients with KD were subdivided into two groups, those with CA sequelae (sCA), such as aneurysms with or without stenosis, and those with no CA involvement (nCA).
Data Acquisition and Measurements
Two-dimensional B-mode, M-mode, and Doppler echocardiography were performed using a Philips iE33 system (Philips Medical Systems, Eindhoven, The Netherlands) or a Vivid-i or Vivid-7 (GE Medical Systems, Louisville, KY), with simultaneous measurement of blood pressure using a commercially available automatic sphygmomanometer. Blood pressure measurements were converted into Z scores according to gender and percentile for height. All data acquisition and measurements were made by two trained and experienced echocardiography technicians (J.T. and D.C.). Interobserver and intraobserver correlations were calculated from a random sample of 29 studies to assess the reproducibility and repeatability of derived BPA.
The following is the methodologic description for measurements specific for accurate reproducibility used in our laboratory. Cine loops of five to seven beats with simultaneous electrocardiographic tracings were digitally recorded. For each measure, the average of four or five cardiac cycles was used as a final representation of that particular measure. A left long-axis parasternal view permitted us to record the aortic root and annulus in B mode and to measure the aortic annulus in systole, from trailing edge to leading edge, and the cross-sectional area of the aortic valve derived from the classic mathematical formula. A modified long-axis view permitted us to record M-mode tracings of the ascending aorta at the level of the right pulmonary artery. The systolic and diastolic diameters of the ascending aorta were then measured from trailing edge to leading edge, the wall thickness being gain dependent. Measurements were made at two specific times of the cardiac cycle, at the beginning of the QRS complex for end-diastole and before the peak of the T wave for the peak dimension in systole. In a standard suprasternal view, the aortic arch was identified and recorded. An ascending aortic pulsed-wave Doppler interrogation was obtained with the sample volume in the center of the aorta at the level of the valve leaflets. Similarly, a Doppler sample recording was obtained at the descending aorta in the same suprasternal view immediately after the acquisition of the ascending aortic Doppler to limit any potential change in heart rate. The peak aortic velocity at the aortic valve level was measured from the average of three or four cardiac cycles. For pulse-wave velocity (PWV) calculation between the ascending and the descending aorta, the time interval between these two points was calculated by subtracting the time between QRS and the onset of the ascending aortic Doppler envelope from that measured between a similar point of the QRS complex and the onset of the descending aortic Doppler envelope. Finally, the aortic arch length was measured between the level of the aortic leaflets and the point of Doppler acquisition in the descending aorta, by summing five linear measurements along the most possible central axis of this curved segment of the aorta in systole ( Figure 1 ).
Method of Calculation of BPA
The noninvasive methodology we used was derived from the Doppler echocardiographic method on the basis of a previously described technique. The required measurements for BPA calculations and the calculation equations are provided in Appendix 1 .
Normalization of BPA with Somatic Growth
Because BPA vary with age, we sought to normalize each parameter for growth. Reference values and Z -score equations were computed with the BPA measurements of the healthy reference subjects (4 months to 17.8 years of age) using a modified version of the method of Royston. Each parameter (dependent variable) was tested against weight, height, and body surface area. Body surface area was estimated using the formula of Haycock et al. Least squares regression was used to fit several regression models, including linear, polynomial (up to the third degree), square root, and power (log-log). The power model was tested only when the distribution of the dependent variable was clearly lognormal. Visual inspection and R 2 values were first used to eliminate models with poorly fitted curves. Outliers were identified visually and using the DFFITS method. Heteroscedasticity was considered present if the residual absolute values modeled with the independent variable displayed a statistically significant positive trend or if either the White test or the Breusch-Pagan test was significant at an α level of .05. When significant heteroscedasticity was detected, it was approached using the method previously described by Altman. Final Z -score equations were selected by taking into account the overall fit of the model, the goodness of the distribution, and its independence from the independent variable. To assess Z -score validity, we used the Anderson-Darling test to measure the goodness of fit of our computed Z score for a normal distribution with a mean of 0 and a standard deviation of 1. We also divided the population into three subgroups according to the tertiles of the independent variable and reassessed the distribution of Z scores for each group to ensure that normal distribution with a mean of 0 and a standard deviation of 1 was preserved across the range of the independent variable. In any case, dependent variables that showed persistent departure from normal distribution regardless of the regression used were not treated with the parametric method described above and are summarized as medians and percentiles.
Normalization of Blood Pressure
Because blood pressure depends on somatic growth, we used previously published equations to calculate height centiles for age and gender and for blood pressure.
General Statistical Analysis
Continuous variables are represented as mean ± SD. Categorical variables are expressed as percentages or numbers. Comparison between patients and controls was done using Student’s t test and analysis of variance for normally distributed variables and using nonparametric (rank-sum) tests for variables not normally distributed. The Jonckheere-Terpstra test was introduced to test for trends between classes. Chi-square and Fisher’s exact tests were applied to compare categorical data. Intraclass correlation coefficients were used to evaluate interobserver and intraobserver reliability. P values < .05 were considered statistically significant. The analyses were done using SPSS version 19 (SPSS, Inc, Chicago, IL) and SAS (SAS Institute Inc, Cary, NC).
Results
Subject Characteristics
Between June 2007 and February 2010, studies were performed on 89 patients with KD ( Table 1 ), among whom 57 met the inclusion criteria. Twelve (21%) had sCA and 45 (79%) had nCA (including no history of CA dilatation or resolved CA dilatation). The overall interval between the acute onset of KD and enrollment was 6.8 ± 5.0 years. At the onset of KD, subjects with sCA had lower serum albumin and higher C-reactive protein and erythrocyte sedimentation rates compared with those with nCA, which is consistent with a higher degree of inflammation in patients with sCA. Controls were 145 healthy subjects aged > 1 year. Participants’ gender distribution was comparable between groups, but patients with KD were relatively older (10.6 ± 5.9 vs 8.7 ± 4.7 years, respectively, P = .010), because of patients with KD with sCA ( Table 2 ). Although absolute values of systolic blood pressure appeared higher in subjects with sCA, statistical comparison performed after Z -score normalization of blood pressure showed lower diastolic blood pressures in patients with sCA and systolic blood pressures comparable with those of controls and patients with KD with nCA ( Table 3 ).
