aUnselected patients group.
bHigh-risk patients of intravenous immunoglobulin resistance group.
IVIG indicates intravenous immunoglobulin.
eFigure 1. The study selection flow diagram
eFigure 2. Meta-regression of days of illness before corticosteroids intervention on Log OR of the effect of corticosteroids on CAA
eFigure 3. Meta-analysis for days of fever after intervention
eFigure 4. Meta-analysis for adverse events
eTable 1. Studies excluded from our meta-analysis after reassessment of full-text articles
eTable 2. Assessment of methodological quality of included studies
eTable 3. Results of meta-regression analysis based on baseline variables
eTable 4. Summary of secondary outcome data of included studies
Customize your JAMA Network experience by selecting one or more topics from the list below.
Chen S, Dong Y, Kiuchi MG, et al. Coronary Artery Complication in Kawasaki Disease and the Importance of Early Intervention : A Systematic Review and Meta-analysis. JAMA Pediatr. 2016;170(12):1156–1163. doi:10.1001/jamapediatrics.2016.2055
What is the effect of corticosteroid therapy in patients with Kawasaki disease in to prevent coronary artery complications?
This systematic review and meta-analysis demonstrates that adjunctive corticosteroid therapy was associated with a significantly lower rate of coronary artery complication compared with intravenous immunoglobulin therapy, particularly among high-risk patients. Meta-regression based on known variables demonstrated that the overall efficacy was negatively correlated with the duration of illness before corticosteroids therapy.
This study highlights the importance of “timing” in treating Kawasaki disease; high-risk patients with Kawasaki disease benefit greatly from a “timely” adjunctive corticosteroid therapy strategy.
The timing and selection of patients with Kawasaki disease for corticosteroid use to prevent coronary artery complications remain controversial.
To evaluate the effect of corticosteroid therapy in KD.
Databases of Medline, The Cochrane Library, and the Clinicaltrials.gov website until July 2015. We used the key words [“Kawasaki disease”] and [“steroid” OR “corticosteroid”] to retrieve potentially relevant studies in the databases of Medline, the Cochrane Library, and the Clinicaltrials.gov website until July 2015. Both English and non-English literature was identified. Titles and abstracts were reviewed by 2 authors (S.C. and Y.D.) to determine suitability for inclusion. Relevant articles were reassessed by reviewing the full text. Discrepancies in study inclusion were resolved by consensus (M.G.K.).
Clinical studies that compared corticosteroids plus intravenous immunoglobulin (IVIG) therapy with IVIG therapy alone in treating patients with KD. Studies either using corticosteroids as initial therapy or as rescue therapy were included.
Data Extraction and Synthesis
Investigators independently extracted the data information. Data were quantitatively synthesized using random-effects analysis.
Main Outcomes and Measures
Rate of coronary artery abnormalities.
Sixteen comparative studies characterizing 2746 patients were analyzed. The duration of illness before corticosteroids therapy was significantly shorter in the initial corticosteroids subset than in the rescue corticosteroids subset. The rate of coronary artery abnormalities was significantly lower in adjunctive corticosteroids therapy than in IVIG therapy (odds ratio [OR], 0.424; 95% CI, 0.270-0.665). Meta-regression based on known variables demonstrated that the overall efficacy was negatively correlated with the duration of illness before corticosteroid therapy (P < .001). Subgroup analysis, including studies using corticosteroids plus IVIG as initial therapy, showed a more advantageous effect than IVIG alone regarding coronary artery abnormality prevention (OR, 0.320; 95% CI, 0.183-0.560), whereas this benefit was not found in a subgroup of studies using corticosteroids as rescue therapy. Further analysis found that patients predicted at baseline to be at high risk of IVIG resistance seemed to obtain the greatest benefit from adjunctive corticosteroid therapy regarding coronary artery abnormality prevention (OR, 0.240; 95% CI, 0.123-0.467). The fever duration was significantly reduced in the corticosteroids group. The favorable effects of corticosteroids were conferred without an increased risk of adverse events.
Conclusions and Relevance
This study highlights the importance of timing to prevent coronary artery complication in treating KD. High-risk patients with KD benefit greatly from a timely and potent adjunctive corticosteroid therapy strategy.
Kawasaki disease (KD), also known as mucocutaneous lymph node syndrome, is an acute systemic vasculitis with unexplained causes, affecting mainly infants and children. Since its first description in Japan in l967 by Tomisaku Kawasaki, KD has been increasingly reported worldwide.1,2
Quiz Ref IDKawasaki disease is a great concern among predisposed children because it is likely to result in grave clinical consequences. Approximately 30% to 50% of patients with KD developed transient coronary artery dilation in the acute stage, and about one-fourth progressed to have serious coronary artery abnormalities (CAA), such as coronary artery aneurysm and ectasia, if left untreated.3,4 Patients with CAA carry a significant risk of coronary thrombus formation, coronary artery stenosis, myocardial infarction, and sudden death even years after the acute illness.4,5 Kawasaki disease has been recognized as one of the most common causes of acquired heart disease in children and thus exerts a significant effect on public health.4
Quiz Ref IDConventional treatment of KD with intravenous immunoglobulin (IVIG) and aspirin is useful in alleviating inflammation and reducing the risk of CAA. However, it was reported that about 20% or in some cases as high as 40% of patients with KD were resistant to IVIG therapy and have a significantly higher risk of developing CAA than nonresistant patients.4,6
Corticosteroids have been used for various kinds of vasculitis and theoretically may also be an effective alternative in treating children with recalcitrant KD. However, a number of subsequent studies evaluating the effect of corticosteroids in KD have yielded conflicting findings.7 Although evidence has suggested a beneficial role of corticosteroids as an initial treatment in KD in preliminary meta-analysis published in 2013,8,9 it remains unclear how best to carry out such a treatment strategy and which patient group will yield the greatest benefit. On the other hand, guidelines recommend that corticosteroids be administered as a second-line rescue therapy for patients who fail to respond to the second dose of IVIG.4 Owing to a lack of randomized trials with adequate power, there is no firm evidence of the efficacy of corticosteroids as rescue therapy in patients with KD.4
Therefore, we sought to comprehensively evaluate the effect of corticosteroids in KD either as initial or rescue therapy by conducting a meta-analysis of all available clinical trials.
This meta-analysis was performed in accordance with the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-analyses Statement.10
The population, intervention, comparison, and outcome approach was used for study inclusion. The population of interest was children diagnosed as having KD.4,11,12 The intervention of interest was treatment using adjunctive corticosteroids either as initial or rescue therapy. Comparisons were made between the corticosteroids group and the conventional therapy (IVIG alone) group. Outcome measurements included the incidence of CAA, duration until defervescence, and adverse events in both groups.
We used the key words [“Kawasaki disease”] and [“steroid” OR “corticosteroid”] to retrieve potentially relevant studies in the databases of Medline, the Cochrane Library, and the Clinicaltrials.gov website until July 2015. Both English and non-English literature was identified. Titles and abstracts were reviewed by 2 authors (S.C. and Y.D.) to determine suitability for inclusion. Relevant articles were reassessed by reviewing the full text. Discrepancies in study inclusion were resolved by consensus (M.G.K.).
According to the Cochrane recommendation (Cochrane Handbook for Systematic Reviews of Interventions, version 5.1.0 [http://handbook.cochrane.org/]), we evaluated study quality with respect to study design, study performance, outcome detection, and result validity. Study characteristics (study purpose, study design, inclusion, and exclusion criteria); participant characteristics (age, sex, race/ethnicity, and severity of illness); information of the intervention (treatment preparation, dose, and duration); and assessment of outcome (method, criteria, incidence, and adverse events) were extracted by 2 investigators (S.C. and Y.D.) from eligible studies.
Quiz Ref IDThe primary outcome of this study was the incidence of CAA, which was defined by the Japanese Ministry of Health criteria11 and coronary artery z score system.13 According to the Japanese Ministry of Health criteria, an abnormal coronary artery is defined as one with internal lumen diameter greater than 3 mm in a child younger than 5 years or greater than 4 mm in a child 5 years or older.11 In the z-score system, a score of more than 2.5 or 3 (selection of the cutoff point depended on the original study design) is used to define coronary artery anomalies.13 Intravenous immunoglobulin resistance (or IVIG no-response) was defined as persistent or recurrent fever lasting (or relapsed within) 24 to 48 hours after the initial IVIG treatment. Two-dimensional echocardiogram was used to detect coronary abnormalities.
The secondary outcome was the comparison for duration of fever after intervention. Adverse events reported by included studies were also pooled and evaluated by our meta-analysis.
The meta-analyses were performed on the basis of the intention-to-treat principle. For studies with more than 1 treatment group, we followed the Cochrane recommendation and pooled relative data into a single group (Cochrane Handbook for Systematic Reviews of Interventions, version 5.1.0 [http://handbook.cochrane.org/]). For continuous outcome measurements reported as median and range, mean and standard variance was estimated by using a simulation formula reported by Hozo et al.14 Statistical difference of outcome measurement between 2 groups was estimated by odds ratios (ORs) for dichotomous variables or standard difference in means for continuous variables, with 2-tailed 95% confidence intervals. With respect to the intrinsic differences of study design, a random-effects model was used to estimate the pooled effects of our meta-analyses.
Heterogeneity among studies was calculated by Q test, and statistic value I2was used to represent the degree of cross-study heterogeneity. A fail-safe N test, Begg regression, and Egger regression were used to estimate the publication bias. Meta-regression was performed to explore risk factors likely introducing potential bias to the overall effects, and the proportion of the studies was assessed by a random-effects regression model. In the meta-regression graph, the circle size reflected the weight each study obtained in this meta-regression. We also used sensitivity (sensitivity analysis was performed by iteratively removing 1 study at a time to confirm that our primary results were not driven by any single study) and subset analyses to test the stability of our meta-analysis. Statistical analyses were performed by using a Review Manager Software package (Revman, version 5.0; The Cochrane Collaboration) and Comprehensive Meta-Analysis software version 2.2 (BioStat Inc). All P values were 2-tailed, and the statistical significance was set at .05.
The study selection process is illustrated in eFigure 1 in the Supplement. A total of 681 articles were retrieved by the preliminary search. After screening titles and abstracts, 660 irrelevant reports, including reviews, commentaries, and case reports, were excluded. In further assessment of the remaining 21 relevant studies, 5 noncomparative studies were excluded (study exclusion list in eTable 1 in the Supplement). Consequently, 16 comparative studies were enrolled in this meta-analysis15-30 (the study by Kobayashi et al26 included “unselected patients group” and “high-risk patients of IVIG resistance group”). The conduction of the included studies was reviewed and approved by their relevant ethics committees.
The baseline characteristics of the 16 studies are shown in Table 1. A total of 2746 cases were involved (861 in the corticosteroids group and 1885 in the IVIG group). Ten studies used corticosteroids as an initial treatment and compared the efficacy of corticosteroids plus IVIG therapy with IVIG therapy,15,17-21,25-28 whereas 6 studies used corticosteroids as a rescue treatment after failure of initial IVIG therapy and compared the efficacy of rescue corticosteroids with additional IVIG therapy among IVIG-resistant patients.16,22-24,29,30 Four studies enrolled patients with KD who were predicted to have high risk of IVIG resistance at baseline.25-28 The Harada score was used to reflect the severity of KD.33 Three scoring systems were used to predict the high risk of IVIG resistance at the onset of the disease.31,32,34 Eleven of 16 studies reported the outcome data in the length of fever,15-21,23,24,27,28 and 6 of 16 studies reported adverse events.17,20,21,27-29 The demographic features of the included studies, including sex, mean age, severity of illness, and duration (days) of illness before intervention, are summarized in Table 1. All patients of the included studies received oral aspirin, and the use of corticosteroids and IVIG in individual studies is detailed in Table 2. The evaluation of methodological quality of the 16 studies is presented in eTable 2 in the Supplement.
Meta-analysis for the overall effect (use of corticosteroids as either rescue or initial therapy) of corticosteroid therapy on CAA is shown in Figure 1. We found that adding corticosteroid therapy was associated with a relative risk reduction of 58% in CAA (OR, 0.424; 95% CI, 0.270-0.665; P < .001; random-effects model), and there was a modest heterogeneity across included studies (P = .10; I2 = 32.9%).
Our sensitivity analysis (using the single-study-removed method) showed a good stability in the overall effect of corticosteroids regarding CAA prevention. Estimation for publication bias indicated a low risk of bias introduced by publication with a statistical index of fail-safe N = 74 (P = .56) and Begg regression test with an intercept of −0.85 (P = .25).
eTable 3 in the Supplement summarizes the results of meta-regression analysis based on known baseline variables including sample size, mean age, sex, and duration of illness before corticosteroid intervention. The regression results showed that the duration of illness before corticosteroid intervention posed a significant risk in affecting the overall effect (slope estimate, 0.34993; 95% CI, 0.15-0.55; P < .001). The regression diagram (eFigure 2 in the Supplement) further illustrated that a longer duration of illness before corticosteroid intervention was negatively associated with the treatment effect of corticosteroids on CAA prevention.
As shown in Figure 2A, subset analysis for studies using corticosteroids plus IVIG as initial therapy strategy (mean [SD] duration of illness in this subset was 4.7 [1.6] days, n = 2363) continued to exhibit a substantial advantage on CAA prevention compared with IVIG therapy alone (OR, 0.320; 95% CI, 0.183-0.560; P < .001; random-effects model). However, subgroup analysis for studies using corticosteroids as a rescue therapy strategy after failure of IVIG treatment (mean [SD] duration of illness in this subset was 7.2 [1.6] days, n = 383) did not show a significant benefit of corticosteroid therapy over additional IVIG therapy (OR, 0.852; 95% CI, 0.467-1.555; P = .60; random-effects model) (Figure 2B).
As shown in Figure 2C, for patients who were predicted to be at high risk of IVIG resistance at baseline, adding corticosteroids to IVIG as initial therapy was associated with a significantly lower risk of CAA development (relative risk reduction of 76%) than IVIG therapy alone (OR, 0.240; 95% CI, 0.123-0.467; P < .001, random-effects model).
The secondary outcomes of the included studies are summarized in eTable 4 in the Supplement. Pooled analysis for duration of fever showed that adjunctive corticosteroid therapy was associated with a more rapid resolution of fever than IVIG alone (mean [SD], 0.66 [1.08] days in the corticosteroids group vs 2.18 [2.55] days in the IVIG group; standard difference in means, −1.804; 95% CI, −2.700 to −0.909; P < .001, random-effects model) (eFigure 4 in the Supplement). Sensitivity analysis further demonstrated good stability of this pooled effect.
Studies that reported and compared the rate of adverse events in both the corticosteroids group and IVIG group were included in this meta-analysis. Adverse events were documented in both groups during treatment and the study follow-up period. Events assessment involved conducting physical examination and laboratory testing. Meta-analysis for adverse events showed no significant difference between the 2 groups (8% in the corticosteroids group vs 7.7% in the IVIG group; OR, 1.312; 95% CI, 0.494-3.485; P = .59, random-effects model) (eFigure 4 in the Supplement). Based on individual study records, most adverse events were transient, and no death was reported in the 16 included studies.
Quiz Ref IDOur study demonstrated that adding corticosteroids to conventional IVIG therapy is associated with reduced risk of CAA compared with IVIG therapy alone for patients with KD. This beneficial effect was only observed when corticosteroids were used as an initial therapy other than a rescue therapy when IVIG treatment failed. By meta-regression analysis, we found that the timing of initiating the corticosteroids therapy during the course of illness significantly influenced the efficacy of corticosteroid on CAA prevention. Subgroup analysis further showed that patients who were determined at baseline to be at high risk of IVIG resistance benefit the greatest from the initial corticosteroid therapy regarding CAA prevention. Quiz Ref IDMoreover, pooled analysis for secondary outcomes showed that adjunctive corticosteroids therapy was associated with a more rapid resolution of fever compared with IVIG therapy alone. Also, pooled analysis for safety assessment showed that there was no significant difference in the rate of adverse events between the corticosteroids group and IVIG group. These results highlight the importance of a timely corticosteroid therapy in addition to IVIG as an initial treatment for high-risk patients with KD.
The most devastating sequela of KD is coronary lesions, which is speculated to be caused by acute systemic inflammation. Therefore, the use of corticosteroids as a strategy to treat KD seems intelligible. Previous studies showed that corticosteroids could improve patients’ clinical conditions or laboratory results15,16 but its efficacy to reduce the risk of CAA remained uncertain. In 2006, Inoue et al20 conducted a multicenter, randomized clinical trial in which 178 unselected children with KD were enrolled, and conventional IVIG therapy was administered with or without the addition of prednisolone. At 1 month of follow-up, patients in the corticosteroids group had a significantly lower incidence of CAA than those in the conventional group, suggesting an advantageous role of adjunctive corticosteroids with respect to CAA prevention.20
However, the results of the study by Inoue et al20 were soon challenged by the 2007 US Pediatric Heart Network study.21 In this trial, patients with KD were randomly assigned to receive primary intravenous methylprednisolone or placebo in addition to conventional IVIG treatment. After 5 weeks of follow-up, there was no between-group difference in the duration of hospitalization and fever, and the rate of CAA once again led clinicians to hesitate to use corticosteroids owing to their uncertain efficacy.
In 2012, Kobayashi et al28 performed the RAISE (Randomised Controlled Trial to Assess Immunoglobulin Plus Steroid Efficacy for Kawasaki Disease) study where 242 high-risk patients with KD were randomly assigned to initial prednisolone plus conventional IVIG or IVIG. A 4-week follow-up investigation exhibited favorable outcomes in the corticosteroids group in terms of lower risk of CAA, lower z score for coronary arteries, and lower incidence of needing second-line therapy. These benefits were not at the cost of increasing the risk of adverse events.28
Potential explanations for these conflicting results mainly included the “timing of therapy” and the selection of patient. For example, the median time of the treatment initiation was 2 days later in the Pediatric Heart Network study21 than in the RAISE study,28 and patients who already had coronary complications at baseline were enrolled in the Pediatric Heart Network study,21 leading to selection bias (shown in Table 1). If the major merit of corticosteroid therapy for KD was to derive from an early suppression of vessel inflammation prior to vascular remodeling, it is possible that any delay in the initiation of therapy would very likely raise the risk of CAA. This “therapy timing window” was corroborated by our meta-regression analysis and subgroup analysis. Another difference in the studies was the use of a single-pulsed dose of methylprednisolone in the Pediatric Heart Network study and 2 weeks of daily IVIG plus oral steroids in the RAISE study. This between-study difference probably highlights that a better clinical outcome is correlated not only with the timing of the treatment but also with the potency of the anti-inflammation therapy.
There are several significant aspects of this study. First, an updated list of 16 clinical studies characterizing 2746 cases was included in this meta-analysis. To our knowledge, this is by far the most comprehensive study of the role of corticosteroids in patients with KD. Second, we found that the duration of illness before intervention (the timing of therapy) was negatively associated with the efficacy of corticosteroids in terms of CAA prevention, emphasizing the importance of “therapy timing.” Third, this meta-analysis was consistent with our previous study by showing that corticosteroids were beneficial in CAA prevention among the overall KD population.9 Furthermore, our subgroup analysis for the first time found that corticosteroids most likely exert a beneficial effect when initiated at the diagnosis of KD rather than after the failure of initial IVIG therapy, when corticosteroids may miss the best timing to take effect.
Should all patients with KD receive corticosteroids as a part of initial treatment, or only a selected group of patients with high risk of poor prognosis? This question seemed to be partly answered by a subgroup analysis of this study, in which patients with KD determined at baseline to have high risk of IVIG resistance were found to show the greatest benefit of corticosteroid therapy regarding coronary sequelae. If so, further questions one might raise are how best to early identify patients who are at high risk for IVIG resistance and what the strong risk factors are for the development of CAA. At least 3 different scoring systems for the prediction of IVIG resistance have been developed in Japanese populations.31,32,34 Despite the useful role of these models in the early identification of high-risk patients in previous trials (positive predictive value ranging from 32% to 59%), the predictive ability of these models, particularly in a non-Japanese population, remained to be optimized. A more efficient diagnosis and treatment system is urgently needed by both clinicians and patients.35
In this study, efforts have been made to control relevant variables by conducting sensitivity analysis and meta-regression on possible confounders, thereby increasing the confidence in the effect estimates of the intervention. However, potential bias could not be completely ruled out owing to the intrinsic differences in the design of included studies. Most included studies were conducted in Japan, where the high awareness of KD may lead to an earlier treatment than in other countries. Whether our results are applicable to other regions remains to be elucidated. The selection of corticosteroids as well as the dosing and the treatment length were not all consistent between studies, and we were not able to conduct specific subgroup analyses owing to a limited number of studies. The current evidence for the role of corticosteroids is mainly based on a short-term observation. Future studies designed with long-term follow-up are therefore needed to provide more data on efficacy and particularly safety assessment.
This systematic review and meta-analysis collected data from a large sample of 16 clinical studies involving 2746 cases to give an updated evaluation of 2 different strategies in treating KD and preventing coronary abnormalities. Corticosteroids combined with IVIG as an initial therapy showed a more protective effect compared with conventional IVIG therapy, and the efficacy was more pronounced in high-risk patients at the initiation of intervention, indicating that an early and aggressive initial anti-inflammation therapy for high-risk patients may be beneficial to improve coronary outcomes. Corticosteroid therapy strategy was also correlated with a more rapid resolution of fever, and the merit of corticosteroids was conferred without an increased risk of adverse events as relative to IVIG therapy. These findings suggest an effective role of corticosteroids in treating KD as an initial therapy strategy among high-risk patients.
Corresponding Author: Shaojie Chen, MD, MM, PhD, Department of Cardiology, Shanghai General Hospital, Shanghai First People’s Hospital, Shanghai Jiao Tong University School of Medicine, Hai Ning Road, No. 100, Hongkou District, Shanghai 200080, China (email@example.com).
Accepted for Publication: June 7, 2016.
Published Online: October 17, 2016. doi:10.1001/jamapediatrics.2016.2055
Author Contributions: Drs Chen and Dong contributed equally to this work. Dr Chen had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Chen, Dong, Kiuchi.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Chen, Dong.
Critical revision of the manuscript for important intellectual content: Chen, Wang, Li, Kiuchi, Ling, Zhou, Wang, Martinek, Püurerfellner, Liu, Krucoff.
Statistical analysis: Chen, Dong, Wang, Li, Kiuchi, Zhou, Wang, Martinek.
Administrative, technical, or material support: Dong, Kiuchi, Ling, Pürerfellner, Liu.
Study supervision: Martinek, Pürerfellner, Liu, Krucoff.
Conflict of Interest Disclosures: None reported.
Additional Contributions: We thank all the participants in this study; Dr Chen acknowledges the Fellowship of the European Society of Cardiology.
Create a personal account or sign in to: