Cardiac biomarkers for early detection of cardiac involvement in children with Kawasaki disease: a cross-sectional study

Kawasaki disease (KD), which is the second most prevalent vasculitis disease in children after IgA vasculitis, can cause serious cardiovascular complications. Early detection of cardiac involvement in KD is an essential part in managing and preventing the cardiac sequels. Although some cardiac biomarkers such as cardiac troponin I (cTnI) and N-terminal pro-brain natriuretic peptide (NT-proBNP) have recently been suggested for early diagnosis of cardiac involvement in patients with KD, their applicability is still unclear. Thus, this study aimed to compare the levels of cTnI and NT-proBNP in KD child patients with or without cardiac involvement. brightness

an excellent power for NT-proBNP (AUC=1.000, p=0.001). Both sensitivity and specificity of NT-proBNP at the optimum cut-off point of 1354 pg/ml were 100 percent.

Conclusion
Unlike the cTnI, the NT-proBNP can serve as an excellent objective test for early detection of cardiac involvement. Therefore, in KD-patients with high levels of NT-proBNP additional therapy and closer follow up should be considered.

Background
Although Kawasaki Disease (KD) is the second most prevalent vasculitis disease in children after IgA vasculitis (IgAV, Henoch-Schönlein purpura), it is known as the main cause of acquired cardiac disease in pediatrics (1)(2)(3). KD is typically a selflimiting diseases and the majority of patients show a complete recovery after the acute phase, but in some cases they disclose serious cardiovascular involvement both in acute and chronic stages (4).
KD is more common among the Asian children under 5 years of age (1), and approximately 15-25 percent of children with untreated KD will develop coronary aneurysms. Although the coronary aneurysms in those children could be reduced by intravenous immunoglobulin (IVIG), five percent of those treated still develop coronary aneurysms (4). Moreover, histopathological findings has reported diffuse myocardial inflammation (5, 6). Also, some studies have reported diastolic dysfunction in KD patients (7,8).
It has been postulated that early diagnosis and management of KD can prevent the further serious complications (9). According to the American Heart Association' guidelines, the diagnosis of classic KD is based on clinical manifestations including a prolonged fever that lasts more than five or four days along with four of five of the principal characteristics including the changes in extremities (erythema of palms and soles or edema of hands and feet in acute phase; periungual peeling of fingers and toes in sub-acute phase), polymorphous exanthema, bilateral bulbar conjunctival injection without exudate, erythema and cracking of lips, strawberry tongue, and/or erythema of oral and pharyngeal mucosa, unilateral cervical lymphadenopathy with diameter greater than 1.5 cm (3,9). In the atypical or incomplete form of KD, patients have prolonged fever with less than four of five principal clinical signs (9).
Furthermore, some laboratory findings such as elevated erythrocyte sedimentation rate, C-reactive protein level, hyponatremia and hypoalbuminemia could be helpful in diagnosis of suspected KD patients. Recently, several biomarkers have been suggested for diagnosis of KD including serum T-helper 1 and T-helper 2 cytokines (interleukin 6 and 20, tumor necrosis factor-α and interferon-γ) and serum Nterminal pro-brain natriuretic peptide (NT-proBNP) (10)(11)(12).
Early detection of cardiac involvement of KD is also important in managing and preventing the cardiac sequels in KD patients (13). Transthoracic echocardiography (TTE) is a diagnostic imaging modality of choice for examining the coronary aneurysms; however, more objective diagnostic tools are required for early detection of cardiac involvement in KD patients (3). It is more than a decade that elevated cardiac troponin I (cTnI) has been considered as a reliable marker of cardiac disorders (14), since it is the most sensitive and specific laboratory test for diagnosing the myocardial infarction (14). It has been demonstrated that NT-proBNP is a proper marker of congestive heart failure and it is elevated in acute myocardial injuries (15,16). Also, it has been indicated that NT-proBNP has a high diagnostic value for cardiac involvement in some diseases such as amyloidosis and Marfan syndrome (17). Their findings were lately confirmed by the study of Adjagba et al. and Jung et al. (20)(21)(22). In contrast, a study by Iwashima et al did not present any significant differences in levels of NT-proBNP between KD patients with and without coronary artery lesions, however, those patients who developed mitral regurgitation showed significant higher levels of NT-proBNP (21). To date, none of cTnI and NT-proBNP biomarkers has been adequately evaluated for their clinical applicability in early diagnosis of cardiac involvement in KD patients. Thus, this study aimed to compare the levels of cTnI and NT-proBNP in KD patients with or without cardiac involvement and the association of them with the severity of cardiac disorders.

Methods
In this cross-sectional study, children with KD who were admitted to the children teaching hospital of Tabriz University of Medical Sciences between April 2015 and April 2018, were consecutively included in the study. Diagnosis of KD was made based on the American Heart Association' criteria for classic KD in children (3). The patients with a pervious history of congenital heart disease, uncorrected structural heart abnormality, dilated or hypertrophic cardiomyopathies, a previous episode of KD or carditis, and the ones who were recently treated with IVIG, and those undergoing or had undergone chemotherapy involving cardio-toxic drugs were excluded from the study. Also, the patients with a final diagnosis of incomplete KD due to its ambiguity were excluded as well.
In included KD patients, the coronary artery involvement was evaluated by a pediatric cardiologist using a TTE (Vivid-3, General Electric Co., USA), and coronary arteries were measured and the Z-values were calculated with a similar method described by Miura et al. (23). Cardiac involvement was defined as the intra-luminal diameter of z-score of ≥2.5 (coronary artery aneurysm) or perivascular brightness of coronary arteries (24). Also, we classified the coronary artery abnormalities as small if the z-score was ≥ 2.5 to <5, large if the z-score was ≥ 5 to < 10, and giant if the z-score was ≥ 10 (25).
Demographic data and KD information were recorded for each child. The onset of KD was defined as the onset of fever in the children. Blood specimens (3 mL from each patient) were obtained at time of admission before any treatment intervention. Data were analyzed using SPSS version 22. Normal numerical data were reported as mean ± standard deviation and un-normal ones were reported as median (minimummaximum). Categorical data were reported as frequency (percentage). Statistical comparisons between the two groups (with or without cardiac involvement) were performed by a two-tailed Student's t-test for numerical data and a Fisher's exact test for categorical data. The power of predicting cardiac involvement was evaluated based on the serum levels of NT-proBNP using the receiver operating characteristic curve (ROC). Also, their areas under the curve (AUCs) and specificity and sensitivity were calculated. A P-value of less than 0.05 was considered as significant. In all study patients, the cTnI levels were lower than 0.35 and the NT-proBNP levels were over than 125 pg/ml exceeding normal ranges. The NT-proBNP measurements revealed a median of 678.5 (220-5735) pg/ml. Also, the highest level of NT-proBNP belonged to the patient who had perivascular brightness of coronary arteries (5735) beside. The other two patients with perivascular brightness of coronary arteries had NT-proBNP levels of 2427 and 5276 pg/ml. The level of NT-proBNP in the patient with small aneurysm of coronary arteries was 1427.

Results
The means of age in the patients with and without cardiac involvement were 2±2 and 3±2, respectively, which had no statistical difference (p=0.343, Table 1). Of those without cardiac involvement, 21 (72.4%) children were male and 8 (27.6%) were female; however, all children with cardiac involvement were male (p=0.001, Table 1). Patients with cardiac involvement had significantly higher levels of NT-proBNP (p=0.001; Figure 1). ROC analysis for power of the NT-proBNP in predicting the cardiac involvement in children with KD, revealed an excellent power for NT-proBNP (AUC=1.000, p=0.001, Figure 2). Both sensitivity and specificity of NT-proBNP at the optimum cut-off point of 1354 pg/ml were 100 percent.

Discussion
The current study investigated the applicability of NT-proBNP and cTnI levels in diagnosing the cardiac involvement in children with KD. Our results demonstrated that NT-proBNP was elevated in KD patients particularly in those with cardiac involvement including perivascular brightness of coronary arteries or coronary artery aneurysm; however, cTnI was in the normal range in all patients regardless of cardiac involvement. NT-proBNP showed an excellent power in predicting the presence of cardiac involvement in KD patients. There were no significant differences difference between the KD patients with and without cardiac involvement in age and the average days between onset of KD and referring to hospital. The majority of patients without cardiac involvement and all patients with cardiac involvement were males. It has been indicated that males are in a higher risk of developing coronary artery aneurysm than females (26).
Our results were in agreement with the pervious findings in value of NT-proBNP in predicting the presence of cardiac involvement (20)(21)(22). However, they had limited the definition of cardiac involvement to the development of coronary artery aneurysm but we also assessed the presence of perivascular brightness of coronary arteries as a preceding factor of developing of coronary artery aneurysm or subsequent coronary ectasia (27,28). Though detecting this abnormality in echocardiography is qualitative and its evaluation depends on the experience of individual echocardiographers. Therefore, an objective test like serum NT-proBNP could be a useful substitute in this regard.
Although Iwashima et al. found no significant difference in levels of the NT-proBNP between those patients with and without coronary artery lesions, they reported that those patients with valvular dysfunction had significantly higher levels of NT-proBNP (21). Also, in our study, the highest level of NT-proBNP was belonged to a patient with TR. It is postulated that TR in KD patients could be a great predictor in their ICU admission (29). Therefore, we expected NT-proBNP levels to be higher in those with more severe cardiac damage. However, considering that the NT-proBNP levels in other patient with sole perivascular brightness of coronary arteries (without TR) were more than the patient with small aneurysm of coronary arteries, we could not conclude a meaningful association between NT-proBNP levels and the severity of cardiac involvement in our study. Although, if we could enroll more KD patients with cardiac involvement, declaring that association could be more accurate.
ProBNP primarily is synthesized and secreted in response to myocyte stretch.
Another source of proBNP can be the intima of coronary arteries (30). Therefore, the possible mechanism behind the increase of NT-proBNP in KD patients with coronary aneurysm could be explained by the micro-damage of intima of dilated coronary arteries by turbulent bloodstream that results in releasing proBNP (31). Also, two other mechanisms has been postulated including local myocardial inflammation with subsequent production of cytokines, stimulate BNP secretion and local areas of ischemia which can affect the pericardium, myocardium, endocardium, and coronary arteries during the acute phase of KD (32,33).
Previously the optimum cut-off points for NT-proBNP levels in for diagnosing the coronary artery aneurysm in KD patients were indicated to be at 1000 pg/ml with a sensitivity of 83% and a specificity of 68% by Kaneko et al. and to be at 1300 pg/ml with 95% sensitivity and 85% specificity by Yoshimura et al. which were both close to the occupied cut-off point in our study (34). However, a very recent study of Jung et al. found a lower cut-off point of 515.4 pg/ml with a sensitivity of 78.26% and a specificity of 61.63% (22). The high values of sensitivity and specificity of NT-proBNP at the cut-off point of 1354 pg/ml in our study could be due to the evaluation of perivascular brightness of coronary arteries as the preceding factor of developing of coronary artery aneurysm. Also, due to a relatively small number of included patients in the study, we may have missed some overlapping patients. Therefore, we should note some limitations of our study. As far as the appropriate and accessible setting for our study was only the children teaching hospital center, only a limited number of patients with KD, even after extending the study duration, were admitted to the study setting. Also, due to the paucity of patients with cardiac involvement, we could not properly evaluate the association between NT-proBNP levels and severity of cardiac involvement. Since that numerous studies have failed to establish the applicability of routine serum tests (e.g. weight blood cells count, creactive protein, erythrocyte sedimentation rate) for diagnosing the cardiac involvement in KD patients, we did not evaluate those serum tests.

Conclusion
NT-proBNP is elevated in KD patients particularly in those with cardiac involvement; however, cTnI does not change in the serum of KD patients. NT-proBNP can detect the presence of cardiac involvement with an excellent power and high rates of sensitivity and specificity. Therefore, in KD patients with high levels of NT-proBNP, complete evaluation, more rigorous therapy, and closer follow-ups should be considered.   Figure 1 Boxplots of NT-proBNP levels in KD patients based on cardiac involvement STROBE_checklist_v4_combined.pdf