Soluble LR11 as a Novel Biomarker in Acute Kawasaki Disease

Abstract

Background: Intimal smooth muscle cells (SMCs) play an important role in the vasculitis caused by Kawasaki disease (KD). Lipoprotein receptor 11 (LR11) is a member of the low-density lipoprotein receptor family, which is expressed markedly in intimal vascular SMCs and secreted in a soluble form (sLR11). sLR11 has been recently identified as a potential vascular lesion biomarker. sLR11 is reportedly elevated in patients with coronary artery lesions long after KD, but there is no description of sLR11 in acute KD. Our aim was to determine the sLR11 dynamics in acute KD and to assess its usefulness as a biomarker.

Methods and Results: 106 acute KD patients and 18 age-matched afebrile controls were enrolled. KD patients were classified into the following subgroups: intravenous immunoglobulin (IVIG) responders (n=85) and non-responders (n=21). Serum sLR11 levels before IVIG therapy were higher in non-responders (median, 19.6 ng/mL; interquartile range [IQR], 13.0–24.9 ng/mL) than in controls (11.9 ng/mL, 10.4–14.9 ng/mL, P<0.01) or responders (14.3 ng/mL, 11.7–16.5 ng/mL, P<0.01). Using a cutoff of >17.5 ng/mL, non-responders to initial IVIG therapy were identified with 66.7% sensitivity and 78.8% specificity.

Conclusions: sLR11 can reflect the state of acute KD and might be a biomarker for patient response to IVIG therapy.

Kawasaki disease (KD) is an acute, self-limited generalized systemic vasculitis that is most common in children under 5 years of age. Although more than 50 years have passed since KD was first reported by Dr. Tomisaku Kawasaki in 1967,¹ its etiology remains unknown. KD causes inflammation in the walls of medium-sized arteries, particularly the coronary artery. Coronary artery lesions (CALs) occur in 20–25% of untreated KD patients, and include aneurysms, stenosis, and occlusion, which are its most critical complications. These critical complications may lead to myocardial infarction or cardiac death.^2,3 KD is the leading cause of acquired heart disease among children in developed countries.⁴

Editorial p 984

Histological results reveal that coronary arteritis begins 6–8 days after onset and progresses to pan-vasculitis on approximately day 10 of acute KD. Aneurysms develop on or around day 12, and are accompanied by destruction of the internal elastic lamina and media.⁵ Smooth muscle cells (SMCs) are observed in the intima in acute KD vasculitis and are considered to contribute to vasculitis progression.⁶

LR11, which is a member of the low-density lipoprotein (LDL) receptor family, is markedly expressed in vascular SMCs that are present in the intima of atherosclerotic lesions. Membrane-bound LR11 and secreted soluble LR11 (sLR11) are associated with SMC migration from the media to the intima,^7,8 suggesting that LR11 plays a key role in the progression of vascular lesions.^9,10

Recently, we showed the utility of sLR11 as a biomarker of vascular lesions, including a long time after KD,^11–15 but there is no description regarding the role of sLR11 in acute KD. Here, we tested the hypothesis that sLR11 reflects vasculitis in acute KD.

Methods

Patients

We enrolled 106 acute KD patients and 18 age-matched afebrile controls from Niigata University Hospital and its 5 affiliated hospitals between April 2009 and December 2012. The afebrile controls were patients with mild arrhythmias or minor congenital heart diseases without hemodynamic abnormalities, including asymptomatic Wolff-Parkinson-White syndrome, rare premature ventricular contractions, and small coronary artery fistula. KD patients were classified into 2 subgroups on the basis of their response to initial high-dose intravenous immunoglobulin (IVIG). Responders were defined by clinical resolution of fever and symptoms within 24 h of initial IVIG therapy. Non-responders were characterized by persistent or recurrent fever after initial IVIG therapy, requiring additional treatment. All patients fulfilled the diagnostic criteria for KD,¹⁶ and were treated with 2 g/kg of IVIG and oral aspirin (30–50 mg/kg/day) as the initial treatment regimen. All non-responders received IVIG (1–2 g/kg) as the second-line treatment, and 5 of them received ulinastatin and 1 received prednisolone in combination with IVIG. There were 5 patients who were refractory to the second-line treatment, and of them, 4 received pulse methylprednisolone (followed by prednisolone in 3 patients), and 1 received a third IVIG dose (2 g/kg) as third-line treatment. One patient received pulse methylprednisolone after the third IVIG administration, and another patient received a third IVIG dose (2 g/kg) after pulse methylprednisolone as fourth-line treatment.

All the KD patients underwent echocardiography to evaluate left ventricular function, valvular insufficiency, and proximal coronary artery dimensions. CALs were defined on the basis of the diagnostic criteria recommended by the Japanese Ministry of Health and Welfare¹⁷ as follows: internal coronary artery diameter ≥3 mm in children aged <5 years; internal diameter of the coronary artery ≥4 mm in children aged ≥5 years; internal diameter of a segment was ≥1.5-fold greater than the adjacent segment, and the lumen of the coronary artery was clearly irregular. The presence of CALs was assessed ≥30 days after onset. This study was approved by the Institution Ethics Review Board of Niigata University Graduate School of Medical and Dental Sciences, and written informed consent was given by all participants.

Laboratory Measurements

Serum samples were obtained at an outpatient clinic from the afebrile controls and KD patients before IVIG therapy (median, 5 days; range, 2–8 days), after IVIG therapy (median, 14 days; range, 9–18 days), and during convalescence (median, 30 days; range, 19–41 days). sLR11 levels were measured in all samples. The white blood cell count, percentage of white blood cells representing neutrophils, platelet count, and serum levels of C-reactive protein (CRP), aspartate aminotransferase (AST), sodium, and bilirubin were measured before IVIG therapy in KD patients, and the Kobayashi score was then calculated. All patients had completed treatment other than oral aspirin by the time they were designated as “after IVIG”. Blood samples were centrifuged after collection, and the supernatant was frozen in polypropylene tubes and stored at −80℃ until use. sLR11 was measured using a sandwich enzyme-linked immunosorbent assay, as previously described.¹⁸

Statistical Analysis

Data in the tables are shown as the median and interquartile range (IQR). Differences in proportions were tested using χ² analysis. The Kruskal-Wallis test followed by the Steel-0Dwass test for multiple comparisons was used to assess the significance of differences among the 3 groups. The Friedman test was used to test significant changes in sLR11 levels over time. A two-way factorial analysis was performed to compare the differences in the logarithm of sLR11 levels between responders and non-responders over time. Receiver operator characteristic (ROC) curve analysis was used to identify a sLR11 cutoff value that would predict non-responders to IVIG therapy. Correlations were determined by Pearson correlation coefficient analysis in KD patients. Differences were considered to be significant when P<0.05. Data analysis was performed using SPSS 22.0 for Windows software (SPSS Inc., Armonk, NY, USA).

Results

Patients

The clinical characteristics of the study population are shown in Table 1. There were no significant differences in age or sex among the 3 groups. The days of blood sampling and initial IVIG therapy were significantly earlier in non-responders (4 and 4.5 days, respectively) than in responders (5 and 5 days, respectively). The non-responders had a significantly higher incidence of CALs than responders (2% vs. 29%, P<0.001). The percentage of neutrophils and AST levels were significantly higher in non-responders than in responders (median; neutrophils, 82.5% vs. 66.0%, P=0.012; AST, 77 vs. 29 IU/L, P=0.002). Platelet count and sodium levels were significantly lower in non-responders than in responders (platelet count, 24.6×10⁴/μL vs. 32.4×10⁴/μL, P=0.009; sodium, 130 vs. 135 mEq/L, P=0.006). The Kobayashi score was significantly higher in non-responders than in responders (6 vs. 3, P=0.005). In addition, CRP levels tended to be higher in non-responders than in responders (8.4 vs. 6.3 mg/dL, P=0.054).

Table 1. Clinical Characteristics of Patients With Acute Kawasaki Disease and Healthy Controls Before IVIG

	Controls (n=18)	Responders (n=85)	Non–responders (n=21)	P values
Age (years)	2.98 (0.44–6.22)	2.34 (0.98–3.66)	2.93 (1.59–4.38)	0.680
Male/female (%)	8/10 (44)	34/51(40)	8/13 (38)	0.917
Day of blood sampling		5 (4–6)	4 (3.25–5)	0.004
Day of initial IVIG		5 (5–6)	4.5 (4–5)	0.016
CAL (%)		2 (2)	6 (29)	<0.001
WBC (×103/μL)		13.4 (11.2–17.9)	13.0 (9.4–15.5)	0.363
Neutrophils (%)		66.0 (53.2–78.1)	82.5 (73.4–85.0)	0.012
Platelet count (×104/μL)		32.4 (27.4–40.8)	24.6 (16.9–26.7)	0.009
CRP (mg/dL)		6.3 (3.2–9.5)	8.4 (6.8–14.1)	0.054
AST (IU/L)		29 (24–50.5)	77 (53.5–817)	0.002
Sodium (mEq/L)		135 (133–137)	130 (129–133.5)	0.006
Bilirubin (mg/dL)		0.55 (0.44–0.8)	1.1 (0.5–1.2)	0.207
Kobayashi score*		3 (1–4)	6 (4–7)	0.005

Values are median, interquartile range (IQR). *Scoring system suggested by Kobayashi et al.²³ AST, aspartate aminotransferase; CAL, coronary artery lesion; IVIG, intravenous immunoglobulin; WBC, white blood cells.

sLR11 Levels Before IVIG Therapy

Serum sLR11 levels before IVIG therapy were significantly higher in all KD patients (median, 14.9 ng/mL; IQR, 11.9–18.8 ng/mL) than in controls (median, 11.9ng/mL; IQR 10.4–14.9 ng/mL, P<0.05; Figure 1A). Serum sLR11 levels before IVIG therapy were significantly higher in non-responders (median, 19.6 ng/mL; IQR 13.0–24.9 ng/mL) than in controls (P<0.01) or responders (median, 14.3ng/mL; IQR 11.7–16.5 ng/mL, P<0.01; Figure 1B). Serum sLR11 levels before IVIG therapy were significantly higher in CAL patients (n=8) (median, 20.9 ng/mL; IQR, 13.6–24.3 ng/mL) than in controls (n=18) (P<0.05). The serum sLR11 levels in CAL patients did not differ significantly from those in non-CAL patients (n=98) (median, 14.5 ng/mL; IQR, 11.9–17.5 ng/mL; Figure 1C). In addition, there was a weak negative correlation between sLR11 level and blood sampling day (P=0.018, r=−0.243).

Figure 1.

(A) Soluble LR11 (sLR11) levels before initial high-dose intravenous immunoglobulin (IVIG) therapy in healthy controls and all Kawasaki disease (KD) patients. Box plots show the median and the first and third quartiles, and the bars show the minimum and maximum values when they are within 1.5-fold of the interquartile range (IQR). (B) sLR11 levels before initial IVIG therapy in healthy controls, IVIG responders, and non-responders. Box plots show the median and the first and third quartiles, and the bars show the minimum and maximum values when they are within 1.5-fold of the IQR. (C) sLR11 levels before IVIG therapy in healthy controls and KD patients with and without coronary artery lesions (CAL, Non-CAL). Box plots show the median and the first and third quartiles, and the bars show the minimum and maximum values when they are within 1.5-fold of the IQR.

Serial Changes in sLR11 Levels

In all KD patients, serum sLR11 levels were significantly elevated after IVIG therapy (23.6 ng/mL, 18.1–29.8 ng/mL, P<0.001), and there was a decreased but sustained higher level than before IVIG therapy in the convalescent stage (18.7 ng/mL, 13.8–23.2 ng/mL, P<0.01 vs. before IVIG) (Figure 2A). Serum sLR11 levels in non-responders to initial IVIG therapy remained higher than those in responders from before IVIG therapy to the convalescent stage (P<0.001; Figure 2B).

Figure 2.

(A) Serial changes in the soluble LR11 (sLR11) levels in acute Kawasaki disease (KD) patients. Box plots show the median and the first and third quartiles, and the bars show the minimum and maximum values when they are within 1.5-fold of the interquartile range. (B) Comparison of the log levels of sLR11 between the responders and non-responders. Data points represent the mean±standard deviation.

Sensitivity and Specificity of sLR11 Levels in Predicting Responsiveness to IVIG Therapy

The relationship between pre-IVIG sLR11 levels and responsiveness to IVIG therapy was examined using ROC curve analysis. The ROC curve and the cutoff value for sLR11 that identified non-responders to the initial IVIG therapy are shown in Figure 3A. The area under the ROC curve was 0.71. Using a cutoff value of >17.5 ng/mL, we identified the non-responders to initial IVIG therapy with a sensitivity of 66.7% and a specificity of 78.8%. Figure 3B shows the ROC curve that compares the responders to non-responders to a second dose of IVIG. The area under the curve was 0.77. Using a cutoff value of >17.6 ng/mL, we identified the non-responders (to a second dose of IVIG) with a sensitivity of 73.3% and a specificity of 79.1%.

Figure 3.

(A) Receiver operating characteristic (ROC) curve of soluble LR11 (sLR11) before the initial intravenous immunoglobulin (IVIG) therapy distinguishes non-responders and responders to initial IVIG therapy. (B) ROC curve of sLR11 before the initial IVIG therapy predicts non-responders to a second round of IVIG therapy.

Correlation Between sLR11 and Other Laboratory Data

Serum sLR11 levels before IVIG therapy showed a positive correlation with AST levels (P<0.0001, r=0.428) and a weak positive correlation with the neutrophil percentage, bilirubin level, and Kobayashi score (P=0.044, r=0.235; P=0.003, r=0.349; and P=0.034, r=0.242, respectively; Table 2).

Table 2. Correlation Between Soluble LR11 Level and Other Laboratory Data

	n	Correlation coefficient	P value
WBC (×103/μL)	84	–	0.537
Neutrophils (%)	73	0.235	0.044
Platelet count (×104/μL)	77	–	0.218
CRP (mg/dL)	84	–	0.437
AST (IU/L)	81	0.428	<0.0001
Sodium (mEq/L)	75	–	0.904
Bilirubin (mg/dL)	70	0.349	0.003
Kobayashi score*	73	0.242	0.034

*Scoring system suggested by Kobayashi et al.²³ AST, aspartate aminotransferase; CRP, C-reactive protein; WBC, white blood cells.

Discussion

The current study results demonstrated that serum sLR11 levels before IVIG therapy in non-responders were higher than in controls or responders, and high levels were sustained until convalescence. Serum sLR11 before IVIG might be a potential biomarker to predict the response to IVIG therapy in KD patients.

We have previously reported that LR11, which is a member of the LDL receptor family, is markedly expressed in intimal SMCs in atherosclerotic lesions but not in medial SMCs.⁸ LR11 plays a key role in the medial-to-intimal migration of SMCs in injured arteries and the development of atherosclerotic lesions.^7–9 In addition, we have recently shown that serum sLR11 is a potential biomarker for evaluating vascular lesions, including atherosclerosis and coronary arteries.^9,11,12,14 We have also reported that serum sLR11 levels in KD patients with CALs were significantly higher than in both patients without CALs and healthy controls. sLR11 could, therefore, be a novel biomarker for vascular lesions long after KD.¹⁵ However, the role of sLR11 in acute KD patients remains unknown.

There are numerous biomarkers for acute KD. Levels of proinflammatory cytokines such as interleukin (IL)-1, IL-6, tumor necrosis factor (TNF)-α, chemokines, and adhesion molecules are elevated in peripheral blood monocyte/macrophages via nuclear transcription factor kappa B.¹⁹ However, there are few reports that focus on biomarkers that identify vascular lesions such as LR11. Tenascin C is produced by vascular SMCs; therefore, circulating tenascin C also could be a potential biomarker for CAL development and IVIG resistance.²⁰

The current study showed that sLR11 increased in the early stage of acute KD, and remained elevated during IVIG treatment through to convalescence. Other biomarker levels in acute KD show a rapid decrease once the fever has subsided. sLR11 has unique dynamics that suggest its production differs from that of other biomarkers. Although this mechanism has not been fully elucidated, we speculate that sLR11 may mainly be derived from intimal SMCs in the acute and convalescent phases of KD. Intimal SMCs are reported to play important roles in KD vasculitis and coronary sequelae. Coronary arteritis begins 6–8 days after onset and progresses to pan-vasculitis on approximately day 10 of acute KD. In KD patients with coronary sequelae, intimal SMCs persist after the convalescent phase. These findings suggest that sLR11, which is derived from intimal SMCs, may peak even after IVIG treatment and remain elevated in the convalescence phase. This is consistent with our previous report that focused on sLR11 long after KD.¹⁵ In addition, LR11 is also reported to be expressed in peripheral leukocytes, especially CD14⁺ monocytes.²¹ sLR11 may be derived from leukocytes, especially in the hyperacute phase before vasculitis.

A high dose of IVIG is now the gold standard therapy for acute KD, but non-responders have a high risk of developing CALs.²² Therefore, it is important to predict the non-responders to IVIG therapy prior to treatment using biomarkers or risk-scoring systems such as the Kobayashi score²³ to indicate which patients will require additional treatment to prevent CALs. However, the effectiveness of risk-scoring systems is not known in other than Japanese children. More reliable predictive systems to identify high-risk patients are required. Because sLR11 was elevated in non-responders to IVIG therapy, it might be a potential biomarker in acute KD. High sLR11 values (≈ >18 ng/mL) might predict patient response to the first- or second-line IVIG therapy.

In conclusion, our results were consistent with the hypothesis that sLR11 reflects vasculitis in acute KD and could serve as a useful marker to predict IVIG resistance. However, sLR11 might be derived from cells other than those in the vasculature, and investigation of other sLR11 sources might improve our understanding of its role as a biomarker as well as the pathophysiology of KD.

Study Limitations

First, this study was an observational study with a small sample size. Further prospective studies with large groups including different ethnic groups and comparison with other biomarkers or risk-scoring systems are required. Second, the timing of blood sampling before IVIG (mostly on the day of treatment or the day before) was not the same for responders and non-responders. However, sLR11 was higher in non-responders, and although the timing of the blood draw was earlier, this might reflect more rapid deterioration of vasculitis in non-responders. This is also consistent with the negative correlation between sLR11 levels and the blood-sampling day. Third, it is desirable to evaluate the coronary arteries using a Z-score during the acute phase to clarify that sLR11 reflects vasculitis in acute KD. However, we did not evaluate the coronary arteries during the acute phase in this study. Fourth, it is necessary to investigate sLR11 levels in patients with inflammatory illnesses other than vasculitis in the future to confirm whether sLR11 is released from cells other than those in the vasculature. Finally, there were differences in treatment between responders and non-responders, including additional treatment that non-responders received. The effect of additional treatment on the sLR11 level is unknown, but we speculate that additional treatments might decrease sLR11 levels in non-responders.

Acknowledgments

This work was supported by Health and Labor Science Research Grants for Translational Research, Japan (H22-001).

Disclosures

The authors declare that there are no conflicts of interest.

IRB Information

This study was approved by the Institution Ethics Review Board at Niigata University Graduate School of Medical and Dental Sciences (reference no. 20101037).

The study procedures followed were in accordance with the Declaration of Helsinki and the ethical standards of the responsible committee on human experimentation (institutional or regional).

Data Availability

The deidentified participant data will be shared upon request to the corresponding author. For any purpose, the data will be shared as Excel files via E-mail.

References

• 1.   Kawasaki T. Acute febrile mucocutaneous syndrome with lymphoid involvement with specific desquamation of the fingers and toes in children. Arerugi 1967; 16: 178–222 (in Japanese).
• 2.   Suzuki A, Kamiya T, Kuwahara N, Ono Y, Kohata T, Takahashi O, et al. Coronary arterial lesions of Kawasaki disease: Cardiac catheterization findings of 1100 cases. Pediatr Cardiol 1986; 7: 3–9.
• 3.   Kato H, Sugimura T, Akagi T, Sato N, Hashino K, Maeno Y, et al. Long-term consequences of Kawasaki disease: A 10- to 21-year follow-up study of 594 patients. Circulation 1996; 94: 1379–1385.
• 4.   Newburger JW, Takahashi M, Burns JC. Kawasaki disease. J Am Coll Cardiol 2016; 67: 1738–1749.
• 5.   Takahashi K, Oharaseki T, Yokouchi Y. Pathogenesis of Kawasaki disease. Clin Exp Immunol 2011; 164(Suppl 1): 20–22.
• 6.   Shimizu C, Oharaseki T, Takahashi K, Kottek A, Franco A, Burns JC. The role of TGF-beta and myofibroblasts in the arteritis of Kawasaki disease. Hum Pathol 2013; 44: 189–198.
• 7.   Zhu Y, Bujo H, Yamazaki H, Ohwaki K, Jiang M, Hirayama S, et al. LR11, an LDL receptor gene family member, is a novel regulator of smooth muscle cell migration. Circ Res 2004; 94: 752–758.
• 8.   Zhu Y, Bujo H, Yamazaki H, Hirayama S, Kanaki T, Takahashi K, et al. Enhanced expression of the LDL receptor family member LR11 increases migration of smooth muscle cells in vitro. Circulation 2002; 105: 1830–1836.
• 9.   Jiang M, Bujo H, Ohwaki K, Unoki H, Yamazaki H, Kanaki T, et al. Ang II-stimulated migration of vascular smooth muscle cells is dependent on LR11 in mice. J Clin Invest 2008; 118: 2733–2746.
• 10.   Ohwaki K, Bujo H, Jiang M, Yamazaki H, Schneider WJ, Saito Y. A secreted soluble form of LR11, specifically expressed in intimal smooth muscle cells, accelerates formation of lipid-laden macrophages. Arterioscler Thromb Vasc Biol 2007; 27: 1050–1056.
• 11.   Takahashi M, Bujo H, Jiang M, Noike H, Saito Y, Shirai K. Enhanced circulating soluble LR11 in patients with coronary organic stenosis. Atherosclerosis 2010; 210: 581–584.
• 12.   Ogita M, Miyauchi K, Dohi T, Tsuboi S, Miyazaki T, Yokoyama T, et al. Increased circulating soluble LR11 in patients with acute coronary syndrome. Clin Chim Acta 2013; 415: 191–194.
• 13.   Ogita M, Miyauchi K, Jiang M, Kasai T, Tsuboi S, Naito R, et al. Circulating soluble LR11, a novel marker of smooth muscle cell proliferation, is enhanced after coronary stenting in response to vascular injury. Atherosclerosis 2014; 237: 374–378.
• 14.   Jin W, Jiang M, Han X, Murano T, Hiruta N, Ebinuma H, et al. Circulating soluble form of LR11, a regulator of smooth muscle cell migration, is a novel marker for intima-media thickness of carotid arteries in type 2 diabetes. Clin Chim Acta 2016; 457: 137–141.
• 15.   Watanabe K, Suzuki H, Jiang M, Haniu H, Numano F, Hoshina S, et al. Soluble LR11 is a novel biomarker for vascular lesions late after Kawasaki disease. Atherosclerosis 2016; 246: 94–97.
• 16.   Ayusawa M, Sonobe T, Uemura S, Ogawa S, Nakamura Y, Kiyosawa N, et al. Revision of diagnostic guidelines for Kawasaki disease (the 5th revised edition). Pediatr Int 2005; 47: 232–234.
• 17.   Research Committee of Kawasaki disease. Report of subcommittee on standardization of diagnostic criteria and reporting of coronary artery lesions in Kawasaki disease. Ministry of Health and Welfare, Tokyo, 1984.
• 18.   Matsuo M, Ebinuma H, Fukamachi I, Jiang M, Bujo H, Saito Y. Development of an immunoassay for the quantification of soluble LR11, a circulating marker of atherosclerosis. Clin Chem 2009; 55: 1801–1808.
• 19.   Matsubara T, Ichiyama T, Furukawa S. Immunological profile of peripheral blood lymphocytes and monocytes/macrophages in Kawasaki disease [Review]. Clin Exp Immunol 2005; 141: 381–387.
• 20.   Okuma Y, Suda K, Nakaoka H, Katsube Y, Mitani Y, Yoshikane Y, et al. Serum tenascin-C as a novel predictor for risk of coronary artery lesion and resistance to intravenous immunoglobulin in Kawasaki disease: A multicenter retrospective study. Circ J 2016; 80: 2376–2381.
• 21.   Sakai S, Nakaseko C, Takeuchi M, Ohwada C, Shimizu N, Tsukamoto S, et al. Circulating soluble LR11/SorLA levels are highly increased and ameliorated by chemotherapy in acute leukemias. Clin Chim Acta 2012; 413: 1542–1548.
• 22.   Burns JC, Capparelli EV, Brown JA, Newburger JW, Glode MP. Intravenous gamma-globulin treatment and retreatment in Kawasaki disease: US/Canadian Kawasaki Syndrome Study Group. Pediatr Infect Dis J 1998; 17: 1144–1148.
• 23.   Kobayashi T, Inoue Y, Takeuchi K, Okada Y, Tamura K, Tomomasa T, et al. Prediction of intravenous immunoglobulin unresponsiveness in patients with Kawasaki disease. Circulation 2006; 113: 2606–2612.